High Frequency Induction Welding &

Post-Welding Heat Treatment of

Steel Pipes

A dissertation submitted for the

degree of Doctor of Philosophy

by Yan Pei
Girton College

Department of Materials Science and Metallurgy

University of Cambridge
June 2011
Pipelines: the world’s energy lifeline
 Preface

This dissertation is submitted for the degree of Doctor of Philosophy at the

University of Cambridge. The work reported was carried out under the su-
pervision of Professor H. K. D. H. Bhadeshia in the Department of Materials
Science and Metallurgy between October 2007 and September 2010. To the
best of my knowledge, this work is original, except where suitable references
are made to previous work or work done in collaboration. Neither this, nor
any substantially similar dissertation has been submitted for any degree,
diploma or qualification at any other university or institution. This disser-
tation does not exceed 60,000 words in length.

Some of the work described herein has been presented in the following pub-
lications:

P. Yan, Ö. E. Güngör, P. Thibaux, and H. K. D. H. Bhadeshia, Crystal-

lographic texture of induction-welded and heat-treated pipeline steel, Ad-
vanced Materials Research, Vol. 89–91 (2010) 651–656.

P. Yan, Ö. E. Güngör, P. Thibaux, and H. K. D. H. Bhadeshia, Induc-

tion welding and heat treatment of steel pipes: evolution of crystallographic
texture detrimental to toughness, Science and Technology of Welding and
Joining, Vol. 15 (2010) 137–141.

Ö. E. Güngör, P. Yan, P. Thibaux, M. Liebeherr, H. K. D. H. Bhadeshia,

and D. Quidort, Investigations into the microstructure–toughness relation in
HFI welded pipes, IPC2010-31372, Proceedings of 8th. International Pipeline
Conference, 27 September–1 October 2010, Calgary, Alberta, Canada.

Yan Pei

June 2011

i
 Acknowledgments

First of all I would like to acknowledge my supervisor, Professor Harry

Bhadeshia, for his enthusiasm and guidance on my research work.
I am indebted to ArcelorMittal Research and Development, Gent, Bel-
gium, who provides financial support, the materials and data for this project.
I would like to express my sincere thanks to Ms. Özlem Esma Güngör, Dr.
Philippe Thibaux, Dr. Martin Liebeherr, Mr. Eric Hivert, and Dr. David
Quidort for their trust in me and inspired discussions about this project.
I would also like to thank Professor Lindsay Greer for the provision of
laboratory facilities in the Department of Materials Science and Metallurgy
at the University of Cambridge, a lot of technical staff in the department for
help on experiments, and all the current and previous members of the Phase
Transformations and Complex Properties research group during these three
years, for making my time in Cambridge more enjoyable with their help and
friendship, especially Mathew for all the constructive suggestions and help
whenever I was not having a clear mind.
At last, I take this opportunity to express my gratitude to my family
and friends for their love and support.

ii
 Abstract

Steel pipelines for the transmission of gas and oil may extend thousands
of kilometres and they represent one of the most sophisticated engineering
achievements of the modern era.
Some of the steel tubes are manufactured from plates which are formed
into an appropriate shape and then seam welded. An efficient and high pro-
ductivity process for achieving this involves high-frequency induction welding
in which the abutting surfaces are forged together following localised heat-
ing. This disrupts the structure of the steel, so the joining operation is
followed immediately by one or more induction heat-treatments, with the
aim of regenerating the properties degraded by the welding operation. It is
found, unfortunately, that the toughness at the location of the weld junction
does not improve as a result of the heat treatment. An investigation of this
phenomenon is the essential aim of this thesis.
The method involved the use of a variety of characterisation techniques
to investigate, for example, the possibility of a high density of non-metallic
particles decorating the junction, the passive role of decarburisation, mi-
crostructural gradients, micromechanical gradients, etc. The breakthrough
came when it was established for the first time that the poor toughness is
due to the existence of a coarse crystallographic grain size (as opposed to a
metallographic grain size), because clusters of grains adopt a similar orienta-
tion in space. This reduces the resistance to cleavage crack propagation. The
effect is not eliminated by heat-treatment because of an “austenite memory
effect” which regenerates the coarse structure on cooling the pipe to ambient
temperature.
The crystallographic grain size is the major effect which embrittles the
weld of the pipelines in sour service environment. When it comes to less so-
phisticated pipelines steels joined by the same process, both crystallographic
texture and inclusions in the weld junction play important roles in causing
low toughness.
In order to propose a method of remedial measures to strengthen the
welds, a modified post-welding heat treatment has been investigated experi-
mentally, which may in the future be implemented on an industrial scale.

iii
Contents

1 Introduction 1

2 Literature Review 3
2.1 Pipe manufacturing process . . . . . . . . . . . . . . . . . . . 3
2.1.1 High frequency induction welding . . . . . . . . . . . . 4
2.1.2 Advantages . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.3 Quality control . . . . . . . . . . . . . . . . . . . . . . 7
2.1.4 Post-welding heat treatment . . . . . . . . . . . . . . . 10
2.2 Pipeline steel . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.1 Metallurgy and microstructure . . . . . . . . . . . . . . 10
2.2.2 Mechanical properties . . . . . . . . . . . . . . . . . . 15
2.3 Impact toughness of the weld . . . . . . . . . . . . . . . . . . 15
2.3.1 Grain size . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.2 Inclusions . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.3 Constituents of the steels . . . . . . . . . . . . . . . . . 20
2.3.4 Composition . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3.5 Crystallographic texture . . . . . . . . . . . . . . . . . 24

3 Experimental Work 31
3.1 Sample preparation . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2 Mechanical tests . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2.1 Tensile . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2.2 Charpy . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2.3 Bend . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2.4 Microhardness . . . . . . . . . . . . . . . . . . . . . . . 35
3.2.5 Nano-indentation . . . . . . . . . . . . . . . . . . . . . 35
3.3 Microstructural observations . . . . . . . . . . . . . . . . . . . 35
3.3.1 Metallography . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.2 Scanning electron microscopy . . . . . . . . . . . . . . 36
3.3.3 Electron probe microanalysis . . . . . . . . . . . . . . . 40

iv
 3.3.4 Transmission electron microscopy . . . . . . . . . . . . 40
3.3.5 X-ray diffrcation . . . . . . . . . . . . . . . . . . . . . 40
3.4 Metallurgical experiments . . . . . . . . . . . . . . . . . . . . 41
3.4.1 Heat treatments . . . . . . . . . . . . . . . . . . . . . . 41
3.4.2 Simulations . . . . . . . . . . . . . . . . . . . . . . . . 41

4 Welds from X65HIC Grade of Steels 43

4.1 Mechanical properties . . . . . . . . . . . . . . . . . . . . . . . 43
4.2 Morphological microstructure . . . . . . . . . . . . . . . . . . 45
4.2.1 As-welded state . . . . . . . . . . . . . . . . . . . . . . 45
4.2.2 After post-welding heat treatment . . . . . . . . . . . . 51
4.3 Crystallographic texture . . . . . . . . . . . . . . . . . . . . . 56
4.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

5 Welds from X65 Grade of Steels 73

5.1 Mechanical properties . . . . . . . . . . . . . . . . . . . . . . . 73
5.1.1 Ductility . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.1.2 Toughness . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.2 Microstructural variance across the weld . . . . . . . . . . . . 76
5.2.1 As-welded state . . . . . . . . . . . . . . . . . . . . . . 77
5.2.2 After post-welding heat treatment . . . . . . . . . . . . 89
5.3 Microhardness profile . . . . . . . . . . . . . . . . . . . . . . . 91
5.4 Inclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.5 Crystallographic analysis . . . . . . . . . . . . . . . . . . . . . 96
5.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

6 Simulations of the High Frequency Induction Welding and

Post-Welding Heat Treatment 110
6.1 Welds from X65HIC pipe . . . . . . . . . . . . . . . . . . . . . 110
6.1.1 Welding process . . . . . . . . . . . . . . . . . . . . . . 110
6.1.2 First cycle of post-welding heat treatment . . . . . . . 112
6.1.3 Modified post-welding heat treatments . . . . . . . . . 113
6.2 Welds from thin-walled X65 pipe . . . . . . . . . . . . . . . . 130
6.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138

7 General Conclusions and Proposed Future Research 140

A Additional supporting evidence for Welds from X65 Pipes 143

v
Chapter 1

Introduction

Pipelines used for the transportation of natural gas and oil, can be fabricated
using high frequency induction welding [1]. There is a variety of steels used
in such manufacturing operations, but this study is focused on a grade of
commercial steel, designated X65, which belongs to the category of strong
steels and is used widely by the pipeline industry [1–3]. The process of pipe-
making involves the mechanical forming of a plate into a cylinder, and then
the joining of the abutting faces by a rapid, localised induction heating and
joining process.
The final joint is quite narrow, but has a central region about 2 mm in
width which may to some extent be decarburised. Yu [4] mentioned that the
carbon concentration at the joint was reduced by as much as 30% relative
to the base metal. The decarburised zone naturally represents a source of
weakness. However, the welding process is then followed immediately by an
induction heat treatment, with the aim of replacing the coarse microstructure
associated with the welding operation with smaller prior austenite grains and
fine ferritic phases. Another aim of this austenisation is to homogenise the
carbon at the weld junction and hence remove the decarburised layer. A
second cycle of post-welding heat treatment is then applied to produce the
final microstructure which is destined for service [5]. Quality control tests
including cross-weld tensile tests and cross-weld Charpy tests are carried out
to establish the suitability for application [3, 5, 6]
High frequency induction welding has been a common industrial pro-
cessing method since 1960s [7]. However, not much information related to this
technique and the welds it produces is available in the academic literature.
In particular, the properties of induction welds, although well-characterised,
are simply not understood. It follows that the methods available to control
the properties are not clear. The work presented in this thesis has been con-

1
ducted with the aim of identifying the problem of low toughness at the weld
junctions when compared against the metal which is not influenced by the
welding process.
To achieve the goal it has been necessary to employ high resolution mi-
crostructural and micromechanical characterisations. The process involves
the systematic validation or dismissal of all the speculative theories that
have been proposed to explain the local mechanical behaviour of the induc-
tion welds. As will be seen later, there were some surprising outcomes and
novel ideas which seem to rationalise observations and reveal the way forward
towards quality welds.

2
Chapter 2

Literature Review

2.1 Pipe manufacturing process

Figure 2.1: Illustration of (a) the high frequency induction welding process [8]
and (b) path of the high frequency current [7]. ‘P’ is the point where welding
occurs.

There are two main types of pipes for industrial use. One is the seam-
less pipe and the other is produced in the form of a pipe by welding. The
most widely used categories of welding methods for pipe-making are gas-
metal/submerged arc welding and electric resistance/induction welding. The
work presented in this thesis is particularly focused on the steel pipes pro-

3
duced by high frequency induction welding. It needs to be mentioned here
that this technology is also suitable for tubing made of other metals for dif-
ferent purposes [8].

2.1.1 High frequency induction welding

High frequency induction welding is a modern manufacturing method, by
which hot-rolled plates of high strength steels can be made into pipes for
long distance gas and oil transportation. Both longitudinal and spiral seams
can be welded by this technology. Figure 2.1a shows an example of the mill
set-up in the welding station for longitudinal seam-welding, where the hot-
rolled steel plates are curled into a tube shape by pressure rolls, the two
abutting edges of the plate are heated up during their passage through the
induction coils, and pressed together by the welding station pressure rolls.
The convergence point, and the separated abutting edges in front of it forms
a V shape. This point is called the ‘Vee’ apex, which defines the onset of
the joint formation that leads to a weld. The high frequency current flows
along the outside surface of the tube and along the edges of the Vee, so
that the electrical circuit is completed, as shown in Fig. 2.1b. This circuit is
formed in the context of three principal features of high frequency induction
heating [7]:

• an induction effect allows the contactless transmission of power to the

workpiece with the aid of an alternating magnetic field. The induction
coil generates this alternating field according to:
!
Pi = k · f · A2W (2.1)

where Pi is the induced power (kW cm−2 ), k is a constant, AW repre-

sents the Ampére-turns per cm of the inductor and f is the frequency
(Hz).

• A skin effect occurs because at high frequencies, electrical currents

and magnetic fields can exist only in a thin layer at the conductor’s
surface [9]. The thin layer is defined as a skin depth, ε, in cm:

"
"
ε = 5030 (2.2)
µf

where " is the specific resistance (Ω cm), and µ is the relative perme-
ability of the material of which the conductor is made [7].

4
 • The proximity effect means that the high-frequency currents always
flow along the path of least resistance. Two currents flowing in opposite
directions on the same material are mutually attracted, as in the Vee [7].
Warren [10] suggested that the position of welding should be within 6
to 14 mm upstream from the centreline of the induction coils, so that
Vee angle is kept within an acceptable range.

Only the heating along the edges of the Vee is useful for welding. The
temperature rise is localized there because of the combination of the skin
and proximity effects. The current flowing at the tube periphery gives rise
to a heat loss which must be kept small. This is achieved by reducing the
resistance at the tube circumference through substantially broadening the
current path. A suitably designed induction coil is essential [7]. The distri-
bution and penetration of heat from the high frequency current are controlled
by the skin and proximity effects together with the frequency of the power
supply and the mill speed, so the welding frequency and mill speed must be
chosen carefully [11].
Although the proximity effect makes the two abutting edges mutually
attract, it can be impeded by [10]:

• roughness on the contacting surfaces;

• an oxide layer or foreign matter;

• a thin layer of absorbed gas on the oxide surface;

• the relative positions of the abutting edges during the introduction of

the high frequency current to achieve heating.

By applying pressure from the rolls, the melted material with relatively high
content of impurities is expelled from the joint. The metal at the junction
flows towards the inner and outer surfaces of the pipes. The quantity expelled
is defined as the difference between the circumference measured at the Vee
angle in front of the rolls and at pipes at the rear of the rolls, which is
usually 1 ∼ 5 mm. The extent of squeezing can affect the weld strength
and toughness [4]. After welding, the rejected materials are mechanically
trimmed off from both the inner and outer surfaces of the pipe.
Certain mill adjustments and roll designs also determine the quality of
the welding. Figure 2.2 shows five different configurations of the pressure
rolls used in the welding station. Some criteria were followed in the design,

5
such as, the fact that the top flange must be large enough in diameter to fully
contain the top edges, and the head rolls located on top should be as thin
as possible since they should not do any edge forming in bending. The top
head rolls can be used to correct only minor edge mismatches [10]. The ideal
disposition of the abutting edges, their melting and two kinds of mismatch
are illustrated in Fig. 2.3. Both vertical (Fig. 2.3c) and angular mismatch
(Fig. 2.3d) result in uneven heating and heat distribution.

Figure 2.2: Illustration of a variety of roll configurations used in the welding

of pipes [10].

2.1.2 Advantages
High frequency welding includes induction and electric resistance welding.
The latter requires electrical contact to heat the weld area, while deterio-
ration of the final microstructure of the welds and pipes by the electrical
contact normally is not a concern for high frequency induction welding.
High frequency welding is more energy-efficient than conventional elec-
trical resistance or induction welding. As the data from the 1980s show, 60%
of the energy from the power line turns into useful heat during the process [8].

6
Figure 2.3: (a) Ideal relationship is to have the two abutting edges paral-
lel and matched in the vertical direction. (b) The electro-magnetic force
ejects hot metal during the high frequency pipe welding. (c) Two edges are
mismatched in the vertical direction. (d) Two edges are mismatched angu-
larly [10].

Compared to submerged arc welding, the induction process does not require
filler metal and hence it has greater productivity [8, 12].
The narrow weld joint and heat affected zone resulting from high fre-
quency induction welding is stronger than the wider weld from many other
welding processes because of the absence of cast structure and the mini-
mization of the distortion of the joining parts [8]. In order to guarantee the
quality of the welded joint, a lot of details need to be controlled. These are
discussed in the following section.

2.1.3 Quality control

There are several parameters which must be controlled at the welding station
in order to make the products of steel tubes/pipes satisfy the requirements
of service under the gas or fluid pressure. Some of these parameters are:
trimmed width of the material entering the mill, circumferential reduction in
the forming stands and welding station, and size and shape of the unwelded
steel entering the welding station. The circumference of the welded and
outer-diameter trimmed pipe must be less than the girth of the unwelded
tube, so as to make sure that a certain amount of material is squeezed out

7
of the weld junction. All these efforts aim to avoid defects due to improper
power adjustment, which can cause a cold, pasty weld, or no weld at all, or
a hot weld with blow hole type voids [10].
The magnitude of heat input is directly reflected in the microstructure,
which can reveal the quality of the weld so as to determine whether the
variables are adjusted properly or not. The main control parameters are [4]:

• fn , fi , fo : width of the weld junction measured at the geometric neutral

line, inside and outside wall of the pipe.

• hn , hi , ho : width of heat-affected zone measured at the corresponding

position as above.

• Standard range: fo ≈ fi ≈ (1.3 ∼ 3) fn , fn = 0.02 ∼ 0.14 mm

• Warning range: fo ≈ fi ≈ (3 ∼ 5) fn , fn = 0.01 ∼ 0.02 mm or fn = 0.14

∼ 0.17 mm

The optimum welding condition for the input power could be determined
experimentally using an electric resistance welding simulator, nondestructive
defect inspection and impact energy measurements. The optimum heat input
range should be re-established whenever the material conditions are changed.
As the contemporary demand for oil and natural gas continues to in-
crease, the principal specifications driving these demands have been [3]:

1. achievement of higher strength grades which are capable of preheat free

welding with cellulosic electrodes;

2. high steel cleanness for resistance to ductile fracture propagation in the

transportation of natural gas, and high integrity of the longitudinal
weld seam;

3. control of centreline segregation levels to ensure weld quality in strong

small-diameter pipes made from centre slit coils.

The requirement for pipes of higher strength and ductility has led to
an increase in manganese contents in the hot rolled sheet steels. However,
a higher manganese content (≥1.2 wt%) and a greater wall-thickness are
likely to induce “penetrator” defects in the welded zone. These defects are
generally classified as:

1. residual FeO-MnO-SiO2 -(Al2 O3 ) oxides left without being squeezed out

from the joint;

8
 2. exposed cracks on the external surface of the pipe due to cavity forma-
tion and hot cracking;

3. blow-holes including oxides.

Penetrator defects become more frequent as the heat input is increased and
the mill speed reduced, and the Mn/Si ratio. The manganese and silicon
content at the welded joint noticeably decreases with the heat coefficient Q,
which is defined as:
EP IP
Q= (2.3)
vt
where EP is the voltage on the plate (kV), IP is the current on the plate (A),
v is pipe welding speed (m/min) and t is the wall-thickness of pipe. This
heat coefficient is generally employed as an index for pipe welding conditions.
This value is higher with a decrease in the mill speed and with an increase
of the welding heat input [13].
Generally speaking, a weld joint in an induction welded pipe may con-
tain defects caused by environmental factors or inappropriate power during
seam welding [14]. A lot of previous work was done to detect and analyse
these defects from the aspect of fatigue properties [15, 16], since they can
initiate fatigue failure. The fatigue crack propagation rate of a welded joint is
generally lower than that of base metal. This may be caused by the hardened
microstructure and distorted fatigue crack propagation path within irregu-
larly arrayed coarse grains in the weld joint [17, 18]. Fine oxides formed
at the weld joint at optimum input power did not lead to a deterioration
of the fatigue propagation rate [19]. Beller and Holden [20] introduced a
nondestructive test method for the detection and classification of defects.
Steel cleanliness requires low sulphur contents to optimise the pipe-
body fracture toughness and avoidance of clustered alumina inclusions to
minimise the occurrence of ultrasonic testing indications in the vicinity of
the longitudinal weld seam. Thus the maximum sulphur content is re-
stricted to 0.005 wt% and the actual level of calcium is typically lowered
from 0.0035 wt% post injection to 0.0008 wt% post vacuum degassing dur-
ing steel-making [3, 21]. For more severe requirements of ‘rich’ gas1 trans-
portation where higher Charpy energy values may be needed, restrictions on
sulphur content of 0.003 wt% may be necessary.
1
Natural gas containing large amount of liquefiable hydrocarbons.

9
2.1.4 Post-welding heat treatment
The main difficulty in welding is the prevention of an abrupt deterioration
of properties as a result of the appearance of structures, which reduce the
resistance to brittle fracture in the heat-affected zone [22]. For this reason,
a two-cycle post-welding heat treatment is used in this study. The weld was
cooled down to room temperature twice during this process. Treiss [5] re-
vealed that on a commercial grade of steel, StE 415.7 TM2 , a single normalis-
ing treatment produces a slight increase in the toughness and a small increase
in the scatter of the measured data. The toughness properties of the parent
metal were realised in the weld area only when this area is water-quenched
after the first normalising and subsequently reheated into the austenite phase
field. This condition requiring a localised heat treatment of the weld without
impairing the whole pipe can only be achieved by induction heating.

2.2 Pipeline steel

2.2.1 Metallurgy and microstructure
High strength pipeline steel falls into the category of high-strength low-alloy
steels and is produced by controlled rolling, in order to achieve the weldability
and toughness demanded by the oil and gas industry [6, 23]. The steel
weldability is defined by the Ito-Besseyo formula for the low carbon contents
(C<0.18 wt%) [24]:
Si Mn + Cr + Cu Ni Mo V
Cequiv = C + + + + + + 5B wt% (2.4)
30 20 60 15 10
Cequiv < 0.4 is an essential requirement of any structural material to be
welded [25], which helps to avoid cold cracking or hydrogen-induced cracking
induced by high hardness and the presence of martensite [24].
Such steels are hot-rolled at elevated temperatures when they have an
austenite crystal structure followed by cooling using air, forced air, water
or mist depending on the preferred cooling rate [26–28]. This production
route is called thermomechanically controlled processing. During cooling the
austenite transforms partially into allotriomorphic ferrite, and the remaining
carbon-enriched austenite transforms into pearlite, as illustrated in Fig. 2.4.
The microstructure of this X65 pipeline steel shows banding of pearlite-rich
and ferritic-rich areas. The volume fraction of the allotriomorphic ferrite was
2
Chemical composition: Fe–(0.04∼0.14)C–0.45Max.Si–(1.00∼1.60)Mn–0.035Max.P–
0.025Max.S wt%

10
stated as 0.93 and the ferrite grains were about 10 µm in size with curved
grain boundaries. The isolated pearlite colonies mostly occurred at boundary
triple junctions or at α/α grain boundaries. The pearlitic cementite was
found to be Mn-rich. There was no Mn segregation at the boundaries between
pearlitic ferrite and pearlitic cementite [29].

(a)

(b)

Figure 2.4: Microstructure of a X65 steel (Fe–0.07C–1.36Mn–0.19Si–0.002S–

0.013P–0.01Ni–0.2Cr–0.04Nb–0.011Al wt%) consisting of ferrite and (a) iso-
lated pearlite colonies; (b) interconnected pearlite colonies [29].

The strength and toughness of the pipeline steel can be related di-
rectly to the ferrite grain size developed as a result of the thermomechanical
treatment. The principal grain refinement mechanism in controlled rolling
is recrystallization of austenite during hot deformation, known as dynamic
recrystallization [23]. This process is affected by the temperature and the
degree of deformation which takes place in each pass during rolling. The

11
austenite microstructure obtained after rolling, and the cooling conditions,
determine the kinetics of the austenite to ferrite transformation. In the ab-
sence of retained strain in the austenite, ferrite predominantly nucleates at
the austenite grain boundaries [30]. Subsequent ferrite growth is usually dis-
cussed as a process governed by carbon diffusion in austenite ahead of the
transformation interface [31, 32]. Experimental results, however, indicate
a solute drag-like effect from alloying elements such as Mn [33, 34]. Fre-
quently, semi-empirical relationships, based on the Avrami equation [35], are
proposed to quantify the effect of chemical composition and austenite grain
size on the transformation kinetics [36, 37].
The ferrite grain size, dα , has been described by Tamura [38], based on
theoretical consideration:
dα ∝ ϕ−0.17 (2.5)

where ϕ is the cooling rate (◦ C s−1 ). The undercooling, which is required to

start the transformation, increases with increasing cooling rate and austen-
ite grain size. The increased undercooling encourages ferrite nucleation. The
Mn solute drag effect at the moving α–γ interface decreases with undercool-
ing [30].
For small austenite grain size and low cooling rate, ferrite nucleation
occurs only at a few preferred sites at the boundary, i.e., at grain corners or
in areas with favourable crystallographic orientation. In the 1980s, Aaron-
son and co–workers performed pioneering studies to shed some light on the
mechanisms of this type of ferrite nucleation [39–42]. The nucleation temper-
ature was estimated from Enomoto and Aaronson’s model [39–41]. However,
an insufficient quench from the isothermal testing temperature encouraged
ferrite nucleation also on the other grain boundary areas, which is consis-
tent with the present findings for continuous cooling conditions. Further,
Enomoto and Aaronson favour a nucleus composition which is associated
with the maximum driving force and a pillbox shape of the nuclei in accor-
dance with minimal interfacial energy. The nucleation sites are at preferred
grain corners first and then grain edges and finally grain surfaces. Growth of
earlier formed nuclei actually determines how much austenite boundary area
remains available for additional nucleation to take place. This fact is used
under mill conditions by employing accelerated cooling to obtain significant
grain refinement in the final ferrite microstructure [30].
The austenite grain size can be suppressed by micro-alloying the steel
with Al, V, Nb and/or Ti. The carbides and carbonitrides of these elements

12
precipitate progressively during the controlled rolling as the temperature
falls. The fine dispersion of particles can not only control the ferrite grain
size but also strengthen the steel. The concentration of the microalloying
elements, the carbon content of the steel and the austenizing temperature of
the process must be optimised to obtain full benefit from the precipitation
and to avoid the formation of undesirable coarse carbides [23, 43].

Figure 2.5: Illustration of different methods of thermomechanically controlled

processing [44]. recr.: recrystallised, MLE: maximum likelihood estimation
of the critical temperture, TM: thermomechanical treatment, ACC: acceler-
ated cooling, DQ: direct quenching, QST: quenching and self-tempering, IC:
intermediate cooling, A: annealing.

Because of the microsegregation of the alloying elements [45], ferrite/

pearlite banding can happen in the pipeline steels. The effect of banding
on the anisotropy of tensile properties is negligible, but it is significant on
the anisotropy of reduction in the area and impact properties [46–48]. It
also makes the steel susceptible to hydrogen-induced cracking [49]. There
are advanced thermomechanically controlled processes designed carefully to
eliminate the banding of pearlite. Examples of various routes of thermome-
chanically controlled processing are shown in Fig. 2.5. Process I is the con-
ventional thermomechanical treatment aiming at reduction of the austenite
grain size and consequently the final ferrite grain size. Further strengthening
can be achieved through finishing in the intercritical temperature range, as

13
shown in process II. With accelerated cooling, designated process III, a mi-
crostructure of very fine polygonal ferrite and acicular ferrite/bainite can be
obtained [44]. By using different combinations and sequences of the steps in
the thermomechanically controlled processing, different microstructures and
properties can be produced from the same chemical composition. Fig. 2.6
compares the microstructures from a thick-walled (31.4 mm) X65 steel plate
manufactured using thermomechanical rolling schedules (a) with interme-
diate cooling only and (b) with accelerated cooling. The former schedule
produced ferrite with bands of pearlite. A maximum grain size of ASTM 10
(standard E112) was reported [44]. The latter resulted in even finer ferrite.

(a) (b)

Figure 2.6: Microstructures of a X65 steel [Fe–(0.04∼0.16)C–0.55Max.Si–

(1.00∼1.60)Mn–0.035Max.P–0.025Max.S wt% according to standard API
(American Petroleum Institue) 5L] applied (a) thermomechanical treatment
and intermediate cooling; and (b) thermomechanical treatment and acceler-
ated cooling [44].

A uniform microstructure in a clean steel has the best resistance to

hydrogen-induced cracking, which is the most common damage mechanism
in pipelines conducting corrosive materials. The hydrogen is produced during
any corrosion reactions taking place within the pipe. The high concentra-
tion of hydrogen sulphide in the transmitted natural gas reacts with iron as
follows:

Fe + H2 S → FeS + 2H, 2H → H2

The atomic hydrogen moves easily within the steel. Cracks begin when the
hydrogen is trapped in the sites of hard phase or inclusions, so the quantity
of non-metallic inclusions in the steel must be minimized in order to avoid
accumulation of the hydrogen. Hydrogen-induced cracking-resistant steels
contain low carbon (optimum range from 0.01 to 0.05 wt% [50]) and low

14
sulphur (as discussed in section 2.1.3). Such steels are designed with inclusion
shape control and limited manganese segregation, since elongated MnS is
found to be detrimental to the hydrogen-induced cracking resistance of the
steel [45, 49, 51, 52].

2.2.2 Mechanical properties

The work presented in this thesis is particularly focused on an API 5L grade
X65 steel and the high frequency welded pipes made from it. Pipeline steels
are designated as Grade A, B and X [53]. X indicates that the steel contains
niobium, vanadium, nitrogen, or other alloying elements. The two digits after
X indicate minimum yield strength requirement of the steel in ksi [50, 54].
The requirements in ISO3183:2007 for the mechanical properties of the steel
are listed in Table 2.1. It is also required that in the bend test the test
Table 2.1: Strength and toughness of X65 welded pipe (According to product
specification level 2 (PSL2) in ISO3183:2007) with outside diameter less than
1422 mm. The yield strength is the value obtained when the sample gets to
0.5% total extension. The absorbed energy required is the result of Charpy
V-notch impact test conducted at 0 ◦ C. The sample size is specified in the
next chapter.

Yield strength Rt0.5 / MPa 450–600

Tensile strength Rm / MPa 535–760
Pipe body
Rt0.5 /Rm max. 0.93
Absorbed energy KV min. / J 40
Tensile strength Rm min. / MPa 535
Weld
Absorbed energy KV min. / J 27

pieces shall not reveal any cracks or ruptures in the base metal, heat-affected
zone or weld junction longer than 3.2 mm or deeper than 12.5% of the wall
thickness. The pipe should withstand a hydrostatic pressure test without
leakage through the weld seam or the pipe body. In the drop weight tear test,
the average shear fracture area should be greater than or equal to 85% at a
test temperature of 0 ◦ C. All of the three tests are specified in ISO3183:2007,
but the latter two methods are not used in this work, since the focus is on
the toughness of the weld.

2.3 Impact toughness of the weld

Impact toughness represents the ability of a material to absorb energy under
impact loading in the presence of a notch [55]. The impact toughness of most

15
structural steels can be described in terms of the transition from ductile to
brittle behaviour as the test temperature decreases. The most widely used
characterisation method for the ductile-to-brittle transition behaviour is the
Charpy V-notch impact test.
The impact toughness of the weld is a complex phenomenon, which
is affected by many microstructural parameters. Generally speaking, a uni-
form and fine microstructure across the weld with the least quantity of in-
clusions on the junction leads to satisfactory toughness. Particular features
in steels, including grain size, various constituents of steel, inclusions, and
crystallographic texture all have influence on the ductile-to-brittle transition
temperature (DBTT). These features exist in the weld junction and affect its
toughness. This section describes the current knowledge about these features
based on available literature.

2.3.1 Grain size

Grain refinement is well-known to lower the DBTT in steels. The relation
between grain size and DBTT originates from the Cottrell model [56], which
describes the occurrence of ductile-to-brittle transition according to the fol-
lowing condition:
Cµγ − 1
σf = σy + ∆σ ≥ d 2 (2.6)
ky
where σf is the fracture stress, σy is the yield strength, d is the grain size, ∆σ
is the increase in yield strength from work-hardening beyond the yield point,
ky is the Hall-Petch slope, µ is the shear modulus, γ is the effective surface
energy of the implied crack, and C is a constant related to stress state and
average ratio of normal to shear stress on the slip plane. The direct effect of
grain size on the DBTT was then described by [57, 58]:
1
DBT T = D ln d 2 (2.7)

where D is a constant.
A large amount of work followed to investigate the dependency of the
transition temperature, on grain size for various kinds of materials. It is
common to plot the transition temperature against the inverse square root
of the grain size as shown in Fig. 2.7. This kind of plot leads to a relationship
of the form [59]:
1
T = B ∗ + Ad− 2 (2.8)
where the transition temperature, T was termed to be DBTT in some ref-
erences [59] or fracture appearance transition temperature (FATT) in some

16
others [60]. A and B ∗ are constants which can be obtained from the graphs
similar to Fig. 2.7 for different steels. d in this relation was defined more
accurately as the effective grain size, which is not always the metallographic
grain size, the packet size of laths in bainite or martensite was one of the ear-
liest discovered exceptions. The toughness of these steels was affected greatly
by the packet size which corresponds to the cleavage unit crack path [61–63].
This is the distance between two nearest high-angle boundaries, since the
crack will tend to ignore the low-angle boundaries [64]. An effective grain is
a packet or cluster of neighbouring grains or units sharing the same or similar
crystallographic orientation, which is also called the crystallographic grain.
The effective grain size should match the facet size observed on the fracture
surface. By applying this criterion, the measurement of effective grain size
on the orientation images was suggested to be taken using the misorienta-
tion threshold between 15 and 18 ◦ for the bainitic steels [65] and 12 ◦ for the
thermomechanical controlled rolled steels [66].

−1
Figure 2.7: Effect of bainite packet size (dB 2 ) on the FATT. Compositions
of the steels are listed in Table 2.2 [60].

2.3.2 Inclusions
The size, distribution and shape of the inclusions are all parameters affect-
ing the toughness of the steels and their welds. Large non-metallic inclusions

17
Table 2.2: Chemical compositions of steels from which the data were plotted
in Fig. 2.7 [60].

Steels C Si Mn P S Cr N O Al
MC25 0.025 0.04 2.04 0.015 0.006 3.0 0.005 0.007 0.005
MCS25 0.028 0.04 2.05 0.015 0.012 3.10 0.005 0.01 0.005
MC50 0.050 0.04 2.07 0.015 0.003 3.06 0.005 0.007 0.005

(>1 µm) within the matrix are internal sites of stress concentration, which
can either act as cleavage crack initiation sites or facilitate the crack propa-
gation [67]. The stress concentration in the case of an axisymmetric loading
can be evaluated by [68]:
2 P
σ = Σm + Σeq + kEP Eeq (2.9)
3
where Σm is the remote hydrostatic applied stress, Σeq is the remote equiv-
alent applied stress, k is a stress concentration factor which depends on the
shape of the inclusion and is given in Table 2.3, EP is the slope of the hard-
P
ening curve, and Eeq is the remote equivalent plastic strain. Thus the stress
concentration is higher on the axial direction of a prolate (needlelike) inclu-
sion and on the radial direction of a oblate inclusion, both of which result in
easier cleavage. The existence of elongated non-metallic inclusions is there-
fore one of the causes for the anisotropy of the impact properties of the steel
after rolling, a subject which is discussed in section 2.3.5. These inclusions
are effective obstacles to crack propagation transverse to the rolling direction
(RD) [69].

Table 2.3: Stress concentration factor associated with various shapes of in-
clusions. s is the ratio of axial to radial dimension [68].

Shape Direction k
2 1 1+2s2
Axial [
3 3 2 log(2s−1)−1
− 1]
Prolate inclusion (s >>1)
1 1 1+2s2
Radial [
2 9 2 log(2s−1)−1
+ 1]
Spherical inclusion Any 1
2 4
Axial (
3 3πs
− 1)
Oblate inclusion (s <<1)
2 10
Radial (
3 3πs
− 1)

The size of inclusions or the size of clusters of inclusions is another

essential parameter to influence the cleavage fracture resistance. Derived

18
from the Griffith’s equation [70], the critical stress required to propagate the
crack from a penny-shaped inclusion of radius c into a ferrite matrix is given
by [67, 71]:
π 2Em γ 1
σf = ( 2
)2 (2.10)
2 π(1 − ν )c
where Em is the Young’s modulus of the matrix, c is the half crack-length,
and ν is the Poisson’s ratio. Because c is proportional to the particle size, this
equation suggests that a material containing larger inclusions is more prone
to cleavage cracking. Similarly a cluster of inclusions acting together as a
big inclusion is detrimental to the toughness of the material. Consequently,
a high content of inclusions weakens the weld by increasing the DBTT and
decreasing the upper shelf energy level, as shown in Fig. 2.8.

Figure 2.8: The effect of inclusions on the weld metal Charpy V-notch tough-
ness [72].

Small inclusions (<1 µm) are of importance in controling the microstruc-

ture in welds and its heat affected zones. During the austenite-to-ferrite
transformation, small oxide and sulphide inclusions concentrated at the austen-
ite grain boundaries restrict the grain growth and result in large surface area
to volume ratio. This increases the possibility of occurrence of the grain
boundary ferrite and reduces the grain size. Furthermore, the inclusions in-
side the austenite grains provide nucleation sites for acicular ferrite, which
is a steel constituent beneficial to the toughness of the weld and will be dis-
cussed in the following section. In summary, a controlled population of small

19
non-metallic inclusions is necessary in welded steels for optimum proper-
ties [73, 74]. Proper dispersion of other second phase particles, i.e., carbides
and nitrides of titanium, vanadium and niobium is beneficial to toughness in
a similar way through promotion of fine grained structure.

2.3.3 Constituents of the steels

Acicular ferrite An increased volume fraction of acicular ferrite results in
a decrease in Charpy V-notch transition temperature, as shown in Fig. 2.9.
It replies on the fine lath size (typically less than 5 µm) of the acicular ferrite
microstructure obtained by rapid cooling.

Figure 2.9: The effect of acicular ferrite content on the 35 J Charpy V-notch
transition temperature of a submerged arc weld metal [75].

Martensite with limited slip possibilities and high yield strength is

generally known as a brittle phase which should be minimised in the weld
joint. The twinned structure of martensite often acts as a crack nucleation
site, but lath martensite with small packet size produced by rapid cooling is
not necessarily detrimental to toughness [68].
Retained austenite in the form of martensite–austenite (M–A) con-
stituents3 can have a strong embrittling effect, which is widely present in
the weld metal and heat-affected zone due to the rapid cooling and possible
segregation of carbon involved during the process [76]. The FATT can be as-
sumed as a linear function of the area fraction of M–A constituents (%M A)
3
An unresolvable mixture of martensite and austenite.

20
in the weld [77]:
F AT T = A + B(%M A) (2.11)
where B is in a range of 3.4–25.9. Considering the M–A constituent as a
second phase in the matrix, its morphological effect should be discussed. It
is found that volume fraction of elongated M–A constituent (aspect ratio
greater than four) influences the toughness of the weld more seriously than
the total content of M–A constituent [77].
Carbides The size of cementite which is the most common form of
carbide in steels plays an important role in affecting the toughness. The fine
cementite precipitates within lower bainite laths can arrest or deviate cracks
and decrease the DBTT [78]. This effect might be even stronger in tempered
martensite because of the multi-directional nature of the cementite. The
cementite formed during tempering is small in size because the fast cooling
rate minimises the diffusion of carbon and hence carbide segregation [68].

Figure 2.10: The effect of carbide thickness on the DBTT [79].

The transition temperature increases dramatically with the thickness of

carbide, tc (µm), as shown in Fig. 2.10. The following equation was proposed
to predict the FATT (◦ C) taking into account the carbide thickness [80]:
1 1
F AT T = 46 + 0.45σp + 131tc2 − 12.7d− 2 (2.12)

where the precipitation strengthening coefficient σp = 1, d is in mm. Car-

bide filament along ferrite grain boundaries, large inter-lath cementite in
upper bainite and cementite lamella as those consisting of pearlite are brittle
and fracture-prone so as to initiate or facilitate the prorogation of cleavage

21
cracks [68, 69]. The DBTT increases with the pearlite content, the pearlite
colony size, p (mm) and the thickness of the cementite lamella, tp (mm). The
increase of interlamellar spacing, S (mm) however, has an opposite effect on
DBTT. Thus the DBTT (◦ C) can be predicted by an empirical equation for
a fully pearlitic structure [69]:
1 1
DBT T = −335 + 5.6S − 2 − 13.3p− 2 + (3.48 × 106 )tp (2.13)

Considering the role of large cementite particles in cleavage cracking, de-

creasing the carbon content leads to improvement on toughness, as shown in
Fig. 2.11.

Figure 2.11: The effect of carbon content on the energy-transition-

temperature curves for steel [81].

The influence of other kinds of carbides on the toughness of steel is

determined by their nature. For instance, M7 C3 precipitates during slow
cooling and grows easily during tempering, which increases the DBTT. On
the other hand, M23 C6 carbides which do not grow to large sizes are beneficial
to toughness [68]. Chromium, tungsten and molybdenum are often found in
this kind of carbides [69]. A dispersion of NbC or VC precipitates in steel re-
tards grain growth and can therefore lead to improved impact toughness [81].

2.3.4 Composition
In addition to the carbon content as discussed in previous section, increas-
ing quantities of oxygen are progressively embrittling, as shown in Fig. 2.12.

22
Oxygen, and in some cases nitrogen, reduces the effective surface energy
for fracture [58]. The role of nitrogen is not straightforward because of its
interaction with other elements as discussed on page 20, but it is generally
considered to be detrimental to toughness [81]. The formation of non-harmful
nitrides (AlN, VN, TiN and etc.) effectively removes the nitrogen from so-
lution, which benefits toughness.

Figure 2.12: The effect of oxygen on the Charpy impact energy of ferrite [58].

The alloying elements in steel detrimental to low-temperature tough-

ness are ferrite stabilizers, while the elements which improve toughness are
austenite stabilizers, as shown in Table 2.4 [58, 69]. Manganese is particularly
beneficial to toughness, the increase of which not only lowers the transition
temperature but raises the upper shelf energy, Fig. 2.13. The most plau-
sible explanation for this is that manganese reduces the thickness of grain
boundary carbides and redistributes them to intragranular sites. Higher so-
lute content of manganese beyond the range shown in Fig. 2.13 increases the
transition temperature again due to the formation of martensitic structure.
Many empirical equations have been developed to describe the transi-
tion temperature based on combinations of effects on impact toughness by
various factors mentioned previously. Among them, the Gladman–Pickering
equation is proposed for ferrite–pearlite microstructured high strength low
alloy steels [69]:
1 1
IT T = −19+44(wt%Si)+700(wt%Nfree ) 2 +2.2(%pearlite)−11.5d− 2 (2.14)

where IT T is the impact transition temperature (◦ C), wt%Nf ree is free ni-
trogen content, %pearlite is the volume percentage of pearlite, and d is in

23
Table 2.4: The effect of alloy elements on the shift of impact transition
temperature (∆ITT) in ferrite–pearlite structure [58].

Solute ∆ITT / ◦ C at%−1

Sn 500
P 400
Cu 40
Mn -50
Si 40
Ni -6
Mo 25
Co -10
Cr 40

Figure 2.13: The effect of manganese on the transition curves of iron [58].

mm.

2.3.5 Crystallographic texture

The impact toughness of the rolled products, most commonly steel plates,
varies with the orientation in the plates. Figure 2.14 shows that the Charpy
specimen has lowest upper shelf energy when the V-notch is oriented towards
the rolling direction of the plate. Attempts were then made to relate the

24
transition temperature to the crystallographic texture of the ferrite. Bramfitt
and Marder [82] defined a texture parameter:

T P = I111 × I110 − 1 (2.15)

where I111 is the relative intensity of {111} in the rolling plane and I110 is the
relative intensity of {110} in the transverse plane. Several researchers [82–84]
reported that the transition temperature of the steels decreased as T P was
increased, which is demonstrated in Fig. 2.15. Morrison [85] however, devel-
oped the relation of transition temperature in the longitudinal testing (the
Charpy sample’s long axis parallel to the rolling direction) to T P showing
the opposite trend, which can be described by [69]:
1
T = 75 − 13d− 2 + 0.63 T P (2.16)

where T is in ◦ C and d is in mm.

Figure 2.14: The effect of specimen orientation on the Charpy transition

curves [81].

Later studies have been undertaken to clarify the influence of preferred

orientations on the impact toughness. In the low temperature regime, where

25
Figure 2.15: The relation of transition temperature at Charpy V-notch en-
ergy of 15 ft-lb (CV15 ) to the texture parameter [83].

the fracture happens exclusively in a brittle mode, the orientation and dis-
tribution of the {100} plane, which is the cleavage plane of body-centred
cubic crystal [86, 87], is the most important factor affecting the toughness.
The smaller angles between the long axis of the Charpy specimen (normal to
the notch plane) and normal to the {100} planes correspond to more brittle
crystals. Increasing the number of {100} planes subject to the maximum
principal stress leads to easy cleavage [88].
As the temperature rises, the fracture mechanism gradually changes
to a mixed plastic-brittle mode and then to a fully plastic mode at room
temperature, when the fracture occurs by the nucleation and coalescence
of microvoids [55]. The more extensive plastic deformation involved in this
process, the higher absorbed energy is needed for fracture. Since the void
nucleation and growth is mainly through glide and generation of pile-ups
on slip planes, the {110}<111> slip system is the main source of plastic
deformation in BCC crystals [89]. Several studies have shown that the upper

26
shelf energy increased with the volume fraction of grains with {110} planes
parallel to the notch plane [88, 90]. Attention may also be paid to the
distributions of other possible slip planes in BCC crystals, i.e. {112} and
{123}, affecting the impact toughness of the steels at room temperature,
because relatively more carbon atoms in solution can fit on the {110} plane
and they act as obstacles to the movement of the dislocations [91].
A mixture of deformation and transformation texture is associated with
induction welding. It is a complicated phenomenon and there is little in-
formation available in the literature. The base metal for the seam-welded
line-pipe has endured the thermomechanically controlled processing as dis-
cussed in section 2.2.1 and the continuous forming process in order to form
the circular shape before welding [92]. Supposing that the final microstruc-
ture of the base metal consists of ferrite and pearlite only, the deformation
texture develops in when the finish-rolling temperature falls into the inter-
critical region where austenite and ferrite coexist. The ferrite grains which
have already formed by that time do not possess random orientations after
the rolling. The ideal texture components for rolling deformation are listed
in Table 2.5. The continuous forming process, Fig. 2.16, is then carried out
by cold rolling, which obviously adds to the deformation texture.
Table 2.5: Main ideal orientations after rolling deformation. The designation
{hkl}<uvw> refers to rolling plane and rolling direction [93].

BCC crystal FCC crystal

Label Orientation Label Orientation
Cube (D) {100}<001> Cube (D) {100}<001>
Rotated cube (RC) {001}<11̄0> Rotated cube (RC) {001}<11̄0>
Goss (G) {011}<100> S (S1 , S2 , S3 , S4 ) {123}<634̄>
Taylor (T1 ) {1̄11}<211> Copper (C1 , C2 ) {112}<111̄>
Taylor (T2 ) {111}<21̄1̄> Brass (B1 , B2 ) {110}<11̄2>
Cross Taylor (CT1 ) {1̄1̄1}<011> Goss (G) {011}<100>
Cross Taylor (CT2 ) {1̄11}<011> α fibre {011}<uvw>

The texture of the base metal is altered locally by welding. During

the high frequency induction welding process, the weld joint is subjected to
compression from both sides by the pressure rolls in order to squeeze out a
certain amount of molten material and form a steady weld. The deformation
at the joint is bound to induce the change of texture, as indicated by the
metal flow that will be evidenced later in Fig. 4.6a and 5.5. Among the
possible texture components, shear texture as listed in Table 2.6, is possibly
predominant due to the characteristics of the deformation.

27
 Figure 2.16: Illustration of the continuous forming process [92].

Table 2.6: Main ideal orientations in simple shear deformation. The desig-
nation {hkl}<uvw> refers to shear plane and shear direction [94, 95].

BCC crystal FCC crystal

Label Crystal orientation Label Crystal orientation
D1 {1̄1̄2}<111> A∗1 {111}<1̄1̄2>
D2 {112̄}<111> A∗2 {111}<112̄>
E {110}<11̄1> A {11̄1}<110>
Ē {1̄1̄0}<1̄11> Ā {1̄11̄}<1̄1̄0>
J {110}<11̄2> B {11̄2}<110>
J¯ {1̄1̄0}<1̄12̄> B̄ {1̄12}<1̄1̄0>
F {110}<001> C {001}<110>

The transformation texture in steel develops as a consequence of the

displacive transformation of plastically deformed austenite. When displacive
transformation occurs, the orientation of martensite or bainite can have 24
possible crystallographic variants. However, when the transformation occurs
under the influence of stress, some variants have a greater probability to form
so as to generate texture, because the shape deformation from austenite to
martensite or bainite complies with the external stress [96]. The transfor-
mation texture adds to the complication of the texture of the weld, since
the displacive transformation products are likely to form in conditions of the
thermal history and stress involved in the high frequency induction welding.

28
If bainitic or martensitic structure forms in the base metal by accelerated
cooling at the end of thermomechanical controlled rolling, it will introduce
transformation texture, because the prior autenite grains have been deformed
during the rolling process.
There is another mechanism which is believed to affect the microstruc-
ture of displacive transformation products. It has been suggested in pub-
lished works [97–99] that two forms of austenite were observed during the
heating of steels, termed ‘globular’ and ‘acicular’. The globular grains are
considered to be the result of normal diffusion controlled growth accompa-
nying the dissolution of cementite, which leads to grain refinement [98]. Aci-
cular austenite nucleation occurs on stabilised retained austenite at bainitic
or martensitic lath boundaries. The acicular austenite was stated to form
by a reverse martensitic reaction, which is governed by a diffusion-less shear
mechanism [98]. Acicular grains within a given prior austenite grain have
a common orientation owing to growth from retained austenite, and they
coalesce on impingement, leading to a reconstitution of the original prior
austenite grain structure, which is termed the austenite memory effect. The
stronger the memory effect, the higher the ratio of acicular/globular austen-
ite [97]. This mechanism may apply to the post-welding heat treatment
process in the current topic of interest. If the prior austenite grain structure
resulting from welding is reconstructed upon austenisation, the heat-treated
weld will inherit the microstructural characteristics including grain size and
crystallographic texture of the as-welded structure.
The presence of both the retained austenite and the stabiliser of the
lath boundaries is necessary to enable the austenite memory effect to oper-
ate [97]. The minimum content of retained austenite needed for the austenite
memory effect to operate is not clear. Kimmins and Gooch [97] found the
evidence of the memory effect in a steel4 with less than 0.8 wt% retained
austenite, heated up to 980 ◦ C at the rate of 4 ◦ C min−1 . However, this steel
did not maintain the austenite memory at the heating rate of 17 ◦ C min−1 .
The higher heating rate results in a larger proportion of the globular nuclei at
a given temperature above Ac1 , because the nucleation rate increases with the
heating rate according to classic theory, whereas the density of the acicular
nuclei resulted from a diffusion-less mechanism may be considered relatively
constant [97, 98]. When the content of retained austenite was 13.9 wt%, the
4
Chemical composition: Fe–0.22C–0.05Al–0.055As–0.005B–1.02Cr–0.02Co–0.19Cu–
0.94Mo–0.21Ni–<0.01Nb–0.36Si–0.51Mn–0.015P–0.013S–0.02Sn–0.1Ti–<0.05W–0.67V
wt%

29
above steel showed a strong memory effect at the heating rate from ∼1000
down to 2 ◦ C min−1 . At very slow heating rates, most retained austenite
transforms into ferrite and carbide, and recovery and recrystallisation occurs
below the Ac1 , which disrupts the oriented growth of acicular austenite. The
presence of residual elements Al, P, Sn and As, and most of the alloying ele-
ments stabilises the martensitic or bainitic lath boundaries and strengthens
the austenite memory, because it encourages the acicular austenite nucleation
and favours a closer Kurdjumov-Sachs orientation relationship between the
acicular austenite and the ferrite from which it forms [97, 98, 100]. Mat-
suda and Okamura discovered that the acicular austenite started to form
above Ac1 , started to coalesce when the temperature was approaching Ac3 ,
and eventually reconstituted the prior austenite grain just above Ac3 . For
austenisation above Ac3 , increasing the austenisation temperature or time at
a given temperature eliminates the memory effect and leads to grain refine-
ment because of the austenite recrystallisation [97].

30
Chapter 3

Experimental Work

3.1 Sample preparation

The high frequency induction welding and annealing process used in the
fabrication of pipes is illustrated in Fig. 3.1, which shows also the length
and separation of the annealing units. The hot-rolled steel strip after edge

This figure removed for copyright reasons

(a)

(b)

Figure 3.1: (a) A sketch map of the production line of high frequency in-
duction (HFI) welding and annealing, Pyr. is the pyrometer for temperature
observation. (b) Specification of the annealing line.

trimming using high-speed cutting is welded during its passage through the
induction coil. There is no external cooling applied to the pipe after welding;
cooling to room temperature before heat–treatment is due mainly to the con-
duction between the narrow heated region and the rest of the pipe. This is
followed by the induction heat treatment. The processing parameters of the
heat treatment applied on the as-welded pipes are listed in Table 3.1. Three
pipes were studied in this work. One of them is made from hydrogen-induced
cracking-resistant steel, X65HIC and the other two are normal API 5L PSL2

31
pipes made from grade X65 strips with different thicknesses. The composi-
tions of the steels are shown in Table 3.2.

Table 3.1: Processing parameters of the post-welding heat treatment.

Steel grade X65HIC X65 X65

Wall thickness / mm 8.8 8.6 14.8
Pipe diameter / mm 508 610 610
Pipe travel speed
Table / m min
removed
−1
22 reasons
for copyright 15 12
Pyr.1 / C
◦
655 661 809
Annealer I Pyr.2 / ◦ C 754 842 982
Pyr.3 / C
◦
960 1054 1070
Quenching Yes Yes Yes
Pyr.4 / C
◦
327 624 813
Pyr.5 / ◦ C 386 761 897
Annealer II
Pyr.6 / ◦ C 452 836 918
Pyr.7 / C
◦
480 922 941

Table 3.2: Compositions of the unwelded steels in wt%

Steel grade X65HIC X65 X65

Wall thickness / mm 8.8 8.6 14.8
C 0.041 0.055 0.064
Table
Mnremoved for copyright
1.1 reasons
1.5 1.5
Si 0.18 0.2 0.21
P 0.013 0.018 0.018
S 0.0007 0.0017 0.0034
Al 0.032 0.023 0.026
Ni 0.018 0.022 0.24
Cr 0.018 0.021 0.02
Cu 0.013 0.014 0.011
Nb 0.054 0.044 0.049
V 0.038 0.0062 0.0067
Ti 0.0019 0.016 0.013
B 0.0002 0.0002 0.0003
Sb 0.0022 <0.0015 <0.0015
Sn 0.0017 0.0009 0.0012
Mo 0.0014 0.0022 0.0019
Ca 0.0024 0.0029 0.003
As – 0.0023 0.0023
N 0.0053 0.0034 0.0042
O – 0.0043 0.0042

Sampling was carried out on the pipe just after welding and after having
gone through the whole process. All the samples were cut normal to the

32
welding direction, which happens to be parallel to the rolling direction, along
the length of the pipes. A pipe segment with the weld is illustrated in
Fig. 3.2a.

Figure 3.2: (a) The orientation of the pipe segment relative to steel processing
directions. (b) Orientation of Charpy specimen.

3.2 Mechanical tests

3.2.1 Tensile
Some of the pipe segments, which obviously have curvature, were flattened to
make tensile specimens. The tensile tests were carried out at room tempera-
ture on strip test pieces made according to API 5L standard. The geometry
is illustrated in Fig. 3.3. The experiments were performed on an INSTRON
hydraulic tensile machine with a capacity of 1200 kN. Cross-weld tensile tests
were conducted following standard ASTM A370:2005 which requires the weld
junction to be placed in the middle of the gauge length. The cross-head speed
at the beginning of the tensile test was 3 mm min−1 , but after 4 % elongation
and to the point of fracture, it was increased to 20 mm min−1 .
Samples from as-welded X65 pipe with 8.6 mm wall-thickness, which
were heat-treated in the Gleeble system, have also been used to carry out
tensile tests. They are in a size of 8.6 × 10 × 30 mm with the weld junction
placed in the middle of the reference length (25 mm), specified by stan-
dard EN10002-1 annex B. The experiments were performed on an electro-
mechanical testing machine Zwick Z250. The cross-head speed at the start

33
of the tensile test was 1.5 mm min−1 , but after 0.6 % elongation and to the
point of fracture, it was increased to 10 mm min−1 .

Figure 3.3: Specification for strip tensile specimen, where t is the wall thick-
ness of the pipe.

3.2.2 Charpy
The Charpy tests were conducted according to standard ISO 148-1 at a
variety of temperatures. The orientation of the test specimen relative to the
pipe is shown in Fig. 3.2b. For the thick-walled X65 pipe, full size specimens
55 × 10 × 10 mm were used. For the X65HIC and thin-walled X65 pipe, 23
size specimens were used, with just 6.7 mm in the normal direction (ND). In
the cross-weld Charpy tests, the V-notch was placed on the weld junction,
and at 1, 2 and 3 mm away from the junction.

3.2.3 Bend
Cross-weld bend tests were performed at room temperature on the pipe seg-
ments, illustrated in Fig. 3.2a, according to standard EN910. For thick-
walled X65 pipe segments, the tests were carried out for the as-welded state
and after post-welding heat treatment. For the other two kinds of pipes, only
samples after post-welding heat treatment were tested. Two testing condi-
tions were used, designated ‘face bend’ and ‘side bend’. In the face-bend
tests, the force was applied in the normal direction. The width of the sample
from thick-walled X65 in the rolling direction was 20 or 40 mm. The samples
from the other two kinds of pipes were all 20 mm wide in the rolling direc-
tion. In the side-bend tests, the force was applied in the rolling direction
along which the samples were 10 mm wide.
All of the above three kinds of tests were carried out in ArcelorMittal
Research and Development Laboratories. Tested samples were then sent to

34
Cambridge for investigation.

3.2.4 Microhardness
Microhardness measurements were carried out across the welds on the as-
etched ND–TD surfaces of the as-welded samples and after post-welding heat
treatment, using a Mitutoyo microhardness tester with a load of 200 gf and
dwell time of 10 s. The metallographic procedures are discussed in section
3.3.1.

3.2.5 Nano-indentation
During nano-indentation, the depth of penetration is recorded as the load
is applied, which enables the measurement of hardness as a function of the
depth. Although the hardness is obtained by dividing the load by the area
of contact as in conventional testing, the area of contact is calculated from
the depth of the penetration and the known values of the angle or the radius
of the indenter, as shown in Fig. 3.4, where A is the contact area, h is the
depth of penetration, θ is the angle between the axis and one surface of the
Berkovich indenter, R and a is the radius and the radius of the contact area
respectively of the spherical indenter.
Nano-hardness was measured using a MTS XP system. A Berkovich
indenter made of diamond was used. Indentations were made on colloidal
silica polished ND–TD surfaces of all three kinds of pipe segments down
to 200 nm. Surface finishing with colloidal silica (0.06 µm) is necessary,
because the roughness of the surface must be of a smaller amplitude than
the impression depth so that it does not affect the calculation of the contact
area.

3.3 Microstructural observations

3.3.1 Metallography
Most of the microstructural observations were carried out on the ND–TD
surfaces of the pipe segments unless otherwise stated. The surfaces were
ground starting with 240-grit silicon carbide grinding paper and finishing
with 2500-grit, and then polished with diamond paste of 6 µm and 1 µm
grade. Some as-polished samples for optical microscopy, scanning electron
microscopy, investigation in the focused ion beam workstation and X-ray

35
 (a) A Berkovich indenter.

(b) A spherical indenter.

Figure 3.4: Geometry of the nano-indenter.

diffractometry were then etched with 2 % nital (2 ml nitric acid with 98

ml methanol). For orientation imaging using the electron back-scattered
diffraction (EBSD) technique, the samples were finish-polished using colloidal
silica (0.06 µm).
The optical microscopes used included a Zeiss Axiotech hi-specification
optical microscope and an Olympus BH microscope, each fitted with a digital
camera. Grain size measurements were made on as-welded samples using the
Hilliard single-circle intercept procedure [101] according to ASTM standard
E112.

3.3.2 Scanning electron microscopy

Besides ordinary imaging by scanning electron microscopy, energy dispersive
X-ray (EDX) analysis and the EBSD technique were also used. These last
two methods respectively take advantage of the X-rays and back-scattered
electrons emitted from the sample during the bombardment of the sample

36
surface with incident electrons.

EDX analysis

The energy dispersive detector tries to collect as many as possible X-rays

emitted from the specimen, so it is usually positioned inside the chamber of
the scanning electron miroscope at a similar distance or angle to the sample
stage as the secondary electron detector. The key component inside the
detector is a small piece of semiconductor. For each incoming X-ray, the
number of electron–hole pairs generated is proportional to the energy of that
X-ray photon detected. When a voltage is applied across the semiconductor,
the magnitude of the current will also be proportional to the energy of the X-
ray. The time during which a current flows between electrodes is referred to as
a pulse, normally less than 1 µs [102]. Each pulse is amplified and registered
on the multichannel analyser, which can distinguish X-ray energies and is
normally attached to a computer. Finally a histogram of energies from all
the X-rays arriving at the detector can be shown on the display.
There is pulse processing speed limit which is decided by the system
hardware, because a pulse can only be dealt with after the previous pulse
is detected, amplified and sorted by the multichannel analyser. During the
time of this process, the detector ignores the coming X-rays and diverts them
to the pulse pile-up rejection circuit. The total time needed for an analysis
includes the live time when the detector is actually counting the pulse and
the dead time when it is rejecting the pulse [102]. Only when an input count
rate is lower than a threshold are most of the incoming pulses satisfactorily
processed. Exceeding the threshold results in a larger fraction of pulses being
rejected. The input count rate depends on the accelerating voltage of the
electron source and the current of the electron beam. The dead time is
displayed as the percentage of pulses rejected in some commercial software
for EDX analysis.
The minimum detectable concentration, M DC can be approximated
as: √
200 b
M DC = √ % (3.1)
(p − b) t
where b is the background count rate, p is the peak count rate and t is the
counting time. By increasing the time of analysis, the M DC can be lowered.
The M DC by using an EDX system is typically around 0.1 wt% [102], but
this figure may need corrections for some or all of the three factors which are
atomic number effect (Z), absorption (A) and fluorescence (F). Use of these

37
corrections is commonly known as ZAF technique. The absorption correction
often has the most significant effect. The magnitude of this correction can be
quite large for detection of light elements with an atomic number less than
that of sodium. This is because a significant portion of the low energy X-
rays get absorbed by the heavier elements in the specimen and the protective
window of the detector. Another limitation applied when measuring the
distribution of an element by the intensity of the X-rays is that the spatial
resolution is around 1 µm [102].
The EDX analysis in this work was carried out on a JEOL 5800LV and
a Camscan MX2600 scanning electron microscope. The latter was equipped
with a field emission gun. Both of them have an Inca-act EDX system.

EBSD technique

Figure 3.5: Formation of the Kikuchi band from electron back-scattered

diffraction.

When an electron beam bombards the specimen, back-scattered elec-

trons are generated by inelastic scattering. These electrons can travel in all

38
possible directions from a thin layer less than 200 nm beneath the sample
surface [103–105]. Those which satisfy the Bragg diffraction condition for a
particular plane show a change in intensity. They are then projected onto an
imaging plane (a phosphor screen) as Kikuchi lines [106]. A pair of Kikuchi
lines called a Kikuchi band, represents a particular family of crystal planes,
as illustrated in Fig. 3.5, the width of the Kikuchi band is related directly
to the lattice parameter of the crystal. All the Kikuchi bands collected on
the screen consist of the electron back-scattered diffraction pattern (EBSP)
from the place of interest where the electron beam strikes the sample. The
intersection of the Kikuchi bands correspond to a zone axis in the crystal.
The EBSP can be indexed after identifying at least two zone axes. It is then
possible to work out the orientation of the crystal relative to the pre-defined
sample axis. Nowadays the indexing process can be fully automated and
operated by a computer. The pattern recognition is done by using a mathe-
matical method, called Hough transformation [107, 108]. Spatially resolved
EBSPs can be recorded to form an orientation map of the surface of the
specimen, which is called orientation imaging (OIM) [109, 110].
The crystal structrual information, mainly the inter-planar spacings
and the angles between the planes, obtained from the EBSP are also used to
search the diffraction database to get the best match. In this way, automated
phase identification can also be achieved. In practice, several possible phases
are selected before starting the OIM, so that the computer will only try
to match the experimental data to these candidate phases, which saves on
calculation time and actually makes the automation feasible. Qualitative
chemistry determination by EDX or wavelength dispersive spectrometry is
also implemented to assist phase identification [111].
Compared to X-ray diffraction, the EBSD technique allows results re-
lating to crystal orientations, e.g. pole figures, to be obtained on a finer
scale. It provides information about the grain distribution in space and the
microtexture of the material, i.e. the texture on a microstructural scale. In
addition, it provides information about much larger areas than those investi-
gated by transmission electron microscopy. Compared to optical microscopy
and scanning electron microscopy, the OIM can reveal the microstructure
with greater clarity, because the grain structure can be defined exclusively
by the degree of misorientation.
The EBSD technique has been used to study pipeline steels [112–114],
with the focus on the microstructure of the steel itself, but not on the weld

39
seams. Interestingly, Venegas et al. [113, 114] studied the role of crystal-
lographic texture in hydrogen induced cracking of API 5L grade X46 steel
(Fe–0.212C–0.037Si–1.334Mn–0.028P–0.032S–0.009Cr wt%). They pointed
out that the crack propagates intergranularly through the high angle bound-
aries and transgranularly on the cleavage plane oriented to favour propaga-
tion. The latter mechanism is also studied in the present work on the fracture
resistance of the welds.
In this study, a EBSD system with HKL Channel 5 software, attached
to the Camscan MX2600 scanning electron microscope was used to carry out
the analysis on both the unaffected materials and the welds. The orientation
images were taken at an operating voltage of 25 kV, a working distance of
30 mm and a tilt angle of 70 ◦ .

3.3.3 Electron probe microanalysis

Further microstructural characterisation includes electron probe microanal-
ysis, which was carried out using a CAMECA SX100 system at an operating
voltage of 15 kV. A beam current of 300 nA was used on X65HIC as-welded
sample while 100 nA on thin X65 as-welded sample. A step size of 0.1 µm
and a dwell time at each point for 60 ms was used to obtain the distribution
maps of manganese and silicon respectively.
The electron microprobe is designated for quantitative chemical anal-
ysis at the microscopic level. It can in principle detect an element with the
atomic number as small as that of boron. The absolute detection limit is
about 10−14 g in an analysing volume of 10 µm3 [115].

3.3.4 Transmission electron microscopy

Transmission electron microscopy (TEM) was carried out by using a JOEL
200CX microscope. A Helios Focused Ion Beam Workstation was used to
prepare thin lamellae from strings in the weld junction of thin X65 as-welded
sample. A beam current of 34 pA was used to finish-thin on both sides of
the specimen.

3.3.5 X-ray diffrcation

A Philips PW1820 X-ray diffraction goniometer with Cu Kα =1.5406 Å
radiation was used for continuous scanning with scan step time of 12.5 s at

40
a step size of 2θ = 0.05 ◦ on the weld junction. The base metal adjacent to
the weld junction was removed.

3.4 Metallurgical experiments

3.4.1 Heat treatments
Two kinds of heat treatments have been done. One is tempering at 400 ◦ C
for 2 h on all three kinds of as-welded samples. This was carried out in a
CARBOLITE tube furnace. The other one is normalisation at 900 ◦ C for half
an hour on the thick-walled X65 as-welded samples. All the samples were
sealed in quartz tubes beforehand, heat-treated in a CARBOLITE chamber
furnace and followed by air cool.

3.4.2 Simulations
For the X65HIC grade, three kinds of experiments were carried out as below.

• Simulation of the induction welding process was done on the base ma-
terial using a THERMECMASTER thermomechanical simulator. As
specified by the system, the sample is machined into a cylinder with
diameter 8 mm and height 12 mm along the rolling direction.

• Simulations of the induction austenisation were done by placing the as-

welded sample into a CARBOLITE chamber furnace at 960 ◦ C, followed
by air cooling or by quenching in brine.

• Experiments of modified post-welding heat treatments consist of an-

nealing at different temperatures followed by tempering at 480 ◦ C.
Three annealing temperatures were chosen, which were 830, 890 and
1000 ◦ C. All the cycles were executed by placing X65HIC as-welded
sample into a furnace with a preset temperature. The samples used
are small weld segments approximately 30 mm long in the transverse
direction (TD) and 1–3 mm thick in the rolling direction. K-type ther-
mocouple was attached to each sample to reveal the thermal history of
the heat treatment.

Modified post-welding heat treatments were also carried out on the thin
X65 as-welded sample using the Gleeble system at ArcelorMittal Research
and Development, Ghent. The temperatures 1054 and 1150 ◦ C were selected
as the peak values of the first cycle, with a holding time of 12, 23 or 27 s,

41
followed by quenching. Peak temperatures of the second cycle was 890, 922
or 950 ◦ C with the choices of holding time among 12, 23 or 38.4 s and followed
by air cooling. Some experiments were stopped half way to investigate the
microstructure of the weld after the first cycle of the heat treatment.

42
Chapter 4

Welds from X65HIC Grade of

Steels

4.1 Mechanical properties

Figure 4.1: Stress-strain curves of high frequency induction welded X65HIC

pipe.

The mechanical properties of the welded pipe were evaluated using

tensile and Charpy impact tests, which are the fundamental requirements to
ensure structural integrity. The tensile test results are shown in Fig. 4.1 and
listed in Table 4.1. The ultimate strength and elongation of the sample after
post-welding heat treatment was approaching that of the base metal, while
the as-welded sample is relatively brittle. The post-welding heat treatment

43
Table 4.1: Critical values of the results from three tensile tests on high
frequency induction welded X65HIC grade pipe.

Ultimate tensile Total

strength/ MPa elongation/ %
Base metal 588 33
As-welded 627 8
Heat-treated 589 27

(a)

(b)

Figure 4.2: Macrographs of the heat-treated X65HIC weld segment after

bend tests (a) 3 samples after face bend. (b) After side bend.

is clearly effective in restoring the ductility of the weld. This was further
confirmed by a U-bend test, in which the heat-treated weld segment remained
integral without any openings on either ND-TD or RD-TD surfaces of the
segment, as shown in Fig. 4.2.
Base metal (‘base’), as-welded and samples after heat treatment were

44
(a) (b)

Figure 4.3: Charpy impact test energy of the X65HIC pipe as a function of
temperature. ‘J’ designates the weld junction, and when used on its own,
the Charpy notch is located at the fusion surface. ‘J+1’ represents the case
where the notch is 1 mm away from the junction. (a) As-welded condition.
(b) Welded and heat-treated.

machined for Charpy testing with the V-notches located on the weld junction,
1 mm and 3 mm away from the weld junction. Each point in Fig. 4.3 is an
average value of three test results. The Charpy energy of the heat-treated
sample is significantly higher than that of the as-welded specimen, when the
V-notches were located 1 or 3 mm away from the joint. However, when
the V-notch was located on the joint, the improvement in toughness due
to heat treatment was not so obvious, with no enhancement for the test
temperature of −40 ◦ C, as shown in Fig. 4.3b. Although the toughness after
heat treatment exceeds the required absorbed energy at 0 ◦ C according to
the ISO standard (section 2.2.2), the increase in the ductile brittle transition
temperature, particularly at the weld junction is still considered as the weak
point of the structure; since these pipelines should be reliable and durable
in service under extreme weather or environmental conditions. This chapter
describes the investigation of the weakness of the weld junction in spite of
heat treatment following welding.

4.2 Morphological microstructure

4.2.1 As-welded state
Figure 4.4 is the picture of the etched surface of the sample in the as-welded
state, in which the weld can be seen clearly. It is categorised into four zones
in this study, which are base metal, thermomechanically-affected zone, heat-
affected zone and weld junction. The microstructure of the four zones was

45
investigated in detail.
The microstructure of the steel unaffected by the welding, consists of
allotriomorphic ferrite and a small amount of pearlite, illustrated in Fig. 4.5.
As discussed in section 2.2.1, the first phase to form during cooling of the ther-
momechanically controlled processing of the pipeline steel is allotriomophic
ferrite which enriches the remaining austenite with carbon. The latter even-
tually transforms into fine pearlite, visible at high resolution, as shown in
the circled region in Figs. 4.5b and c. The small fraction of pearlite is due
to the low average carbon concentration (0.041 wt%) of the steel. This kind
of microstructure is referred to as ‘base metal’.
During the welding process, the abutting edges of the pipe are pushed
together whilst they are hot and plastic, thus giving rise to flow which
serves two purposes: first, to expel undesirable oxides or molten matters
if there is any from the weld junction, and secondly to form a metallurgical
bond by breaking up the interface between the edges. There is therefore a
thermomechanically-affected zone where remnants of the deformation asso-

Figure 4.4: The four zones which define the X65HIC welded region: 1–base
metal, 2–thermomechanically-affected zone, 3–heat-affected zone, and 4–weld
junction. The arrows illustrate the directions of metal flow.

46
 (a)

(b) (c)

Figure 4.5: Microstructure of the unwelded X65HIC steel in pipe form: (a)
optical micrograph, (b) and (c) scanning electron micrograghs.

ciated with the welding process are retained. This has a structure similar
to that of the base metal, except that the shapes of the ferrite grains are
anisotropic, aligned along the direction of metal flow, Fig. 4.6a. Four parts
of the thermomechanically-affected zones located on both sides of the weld
junction are pointed out in Fig. 4.4. There is no obvious thermomechanically-
affected zone at the mid-thickness of the pipe. It is possible that the align-
ment observed is purely due to the deformation of ferrite, but it could also
be true that the morphology is a consequence of the severe pancaking of the
austenite [38, 116] before it transforms into ferrite.
The heat-affected zone on the as-welded sample consists mainly of
equiaxed allotriomorphic ferrite which has reformed from the grains which
become austenitic and then dynamically recrystallise, Fig. 4.6b. The gra-
dients of microstructure expected in the wide heat-affected zone of conven-
tional fusion welds, i.e., tempered plate, intercritically heated plate and fully
austenitised regions, are not obvious, partly because of the narrow dimen-

47
 (a)

(b)

(c)

Figure 4.6: Microstructure of (a) thermomechanically-affected zone, (b) heat-

affected zone and (c) weld junction on as-welded sample near the outer sur-
face of the X65HIC pipe.

48
sions of the zone but also because the steel has such a low solute-content
and hence, hardenability. Grain size measurements on the heat-affected zone
gave an average value of 5 ± 0.5 µm, while the measurements on base metal
gave 7 ± 0.6 µm.
The microstructure at the weld junction of the as-welded sample is in-
teresting, consisting of a mixture of grain boundary allotriomorphs of ferrite,
Widmanstätten ferrite and an intervening residual phase which is likely to
be M–A islands, Fig. 4.6c. This is a direct consequence of the fact that the
austenite grains here are expected to be coarser, since this region was heated
up most effectively during the welding process. It is well-known that coarse
austenite grains favour the formation of Widmanstätten ferrite [117–119].
The M–A islands etched dark as shown in the optical micrograph, Fig. 4.6c.
They can be seen more clearly in the scanning electron micrograph, Fig. 4.7a.
The X–ray diffraction pattern of the X65HIC as-welded sample shows the ex-
istence of small amount of austenite, as only two tiny austenite peaks can be
identified in Fig. 4.8, which should be contributed by these M–A islands in
the weld junction. After tempering the as-welded sample at 400 ◦ C for 2 h,
the M–A islands turned into tempered martensite, as shown in Fig. 4.7b.
At the mid-thickness of the pipe, several elongated M–A islands could
be observed in the weld junction, as shown in Fig. 4.9a, but the strings were
not present in other parts of the weld. Electron probe microanalysis demon-
strated that they are manganese-rich phases, Fig. 4.10. After tempering
they revealed the classical morphology of tempered marteniste, Fig. 4.9b.
The phenomenon of strings will be further discussed in the next chapter,
since it is more prevalent in the welds from X65 steels.
There are no sharp boundaries between the four different zones across
the weld. The microstructure changes gradually. The micro-hardness profile
across the weld on the as-welded sample is shown in Fig. 4.11a. The origin of
the x-axis corresponds to the centre of the weld junction. Consistent with the
tensile data (Fig. 4.1), the weld is harder than the base metal, especially near
the outer surface of the pipe. The higher hardness near the outer surface of
the pipe, compared to the other half, may be caused by the surface hardening
effect of the induction heating. Although the microstructure of the outer
layer of the pipe is not distinguishable from that of the rest, the induction
heating should leave a compressive residual stress on the outer layer of the
work piece [120, 121], because the surface is cooled down more quickly than
the inner part.

49
(a)

(b)

Figure 4.7: Scanning electron micrograghs of weld junction near the outer
surface of the X65HIC pipe. (a) As-welded state. (b) After tempering at
400 ◦ C for 2 h.

The hardness across the weld is scattered in the mid-thickness and near
the inner surface of the pipe. The peak hardness value falls into the heat-
affected zone on both sides of the weld junction near the inner surface of the
pipe, since this zone is featured with the finer grain structure as mentioned

50
 Figure 4.8: X-ray diffraction pattern for the X65HIC weld.

above. Such an increase of hardness at the mid-thickness should also be

expected. However the weld is much narrower at the mid-thickness of the
pipe, approximately 2 mm in width (Fig. 4.4); so the step size, 150 µm, in
establishing the micro-hardness trace may not be fine enough to reveal the
true distribution of hardness among different zones.
After tempering, the hardness of the weld dropped by 10 to 20 HV,
compared with the as–welded state (cf. Fig. 4.11a and b), which also supports
the existence of martensite in the as-welded sample.

4.2.2 After post-welding heat treatment

The induction post-welding heat treatment broadens the affected area from
the welding process itself, as shown in Fig. 4.12. However, the microstructure
across the heat-treated weld is much more uniform compared to the as-welded
sample. Three microstructural domains could be recognized under extensive
etching with 2 % nital, Fig. 4.12. The affected area has a microstructure
similar to that of the unaffected base metal, consisting of allotriomorphic
ferrite and pearlite (cf. Fig. 4.13a and Fig. 4.5a). The only difference is that
the ferrite grains are more equiaxed in the affected area. The alignment of
the grains due to rolling of the steel sheet in the base metal and forging in
the former thermomechanically-affected zone has been annealed out.
Vestiges of the weld junction remained after heat treatment, since plate-
like ferrite exists in the microstructure, Fig. 4.13b. Scanning electron mi-
croscopy shows evidence of decarburisation; since there is some divorced

51
(a)

(b)

Figure 4.9: Scanning electron micrograghs of the weld junction at the mid-
thickness of the X65HIC pipe. (a) As-welded state. (b) After tempering at
400 ◦ C for 2 hours.

pearlite in the position of the original weld junction, Fig. 4.13d. The af-
fected area away from the weld junction in comparison has more pearlite,
Fig. 4.13c. This amount of decarburisation does not have significant effect

52
 (a)

(b)

Figure 4.10: The weld junction on the as-welded X65HIC sample at the mid-
thickness of the pipe. (a) Scanning electron microgragh. (b) Mn mapping
of (a) using electron probe microanlaysis. This analysis can not be fully
quantified due to experimental limitation. Bigger number and warmer colour
means higher concentration of the elements and vice versa.

53
(a) (b)

(c)

Figure 4.11: Micro-hardness distribution across the weld of the X65HIC pipe.
(a) As-welded state. (b) After tempering at 400 ◦ C for 2 hours. (c) Welded
and heat-treated.

on the mechanical properties of the weld, since the hardness across the weld
after post-welding heat treatment is uniform as shown in Fig. 4.11c, which
is consistent with the unification of the microstructure.

Figure 4.12: Weld segment from the X65HIC pipe after post-welding heat
treatment. 1–base metal, 2–the affected area, and 3–the original location of
the weld junction.

54
(a) (b)

(c) (d)

Figure 4.13: Microstructure of the sample after post-welding heat treatment

near the outer surface of the X65HIC pipe. (a) Affected area. (b) The
original location of the weld junction. (c) and (d) Corresponding scanning
electron micrographs taken at the region of (a) and (b).

The microstructural observation did not reveal any remaining austenite

in the sample after post-welding heat treatment and the X–ray diffraction
pattern in Fig. 4.8 confirms its absence.
The possibility of oxides at the weld junction being the cause of poor
toughness was investigated using extensive optical and scanning electron mi-
croscopy. These experiments did not reveal any role of inclusions in deter-
mining the impact properties at the junction. Broken Charpy samples from
the weld junction were also examined because there is a better chance of
detecting significant initiating features, but this led to the same conclusion.
A typical fracture surface of the sample after post-welding heat treatment,
tested at the weld junction at −40◦ C, is shown in Fig. 4.14.
By this point of the investigation, it was concluded that the post-
welding heat treatment is effective in removal of any microstructure detri-
mental to the mechanical properties of the regions influenced by the weld-
ing. All the conventional causes of the low toughness seemed to be excluded
from consideration: there was neither extraordinarily coarse microstructure

55
Figure 4.14: Fracture surface at weld junction of the X65HIC pipe after
post–weld heat treatment, tested at −40◦ C.

nor substantial decarburisation in the weld junction, no retained austenite

was detected, and fractographic analysis did not reveal evidence of inclusion-
induced fracture. However, there was a breakthrough, as revealed in the next
section, where the reason for low toughness at the weld junction is shown to
be crystallographic in origin.

4.3 Crystallographic texture

A large quantity of crystallographic data were collected using EBSD in order
to give a complete overview of the texture of the weld. Ferrite orientation
images were constructed from 2 mm away towards the weld junction near the
outer (Fig. 4.16) and inner (Fig. 4.18) surfaces of the pipe, and from 1 mm
away at the mid-thickness as shown in Fig. 4.17 again because of the narrower
width of the weld at the mid-thickness of the pipe. Figure 4.15 shows the
colour coding of the orientation images in this study unless otherwise stated.

It is noted here that austenite was not confidently detected by EBSD

scanning even in the as-welded sample, because the austenite exists in minute
quantities associated with the M–A islands, the hybrid microstructure of
which makes it difficult for recognition of the EBSP. In all three images, the
scan starts 2 mm away from the weld junction at the top left and contin-

56
Figure 4.15: Colour codes in the orientation images, representing the inverse
pole figure information projecting on the plane normal to rolling direction.

ues towards the middle of the weld junction at the bottom right. On the
scanned sample, 2 mm away from the weld junction is the base metal near
the outer surface of the pipe, so the microstructure variance across the weld
at the as-welded state is shown clearly, Fig. 4.16. Grain boundaries in the
orientation images are designated mainly by the solid black lines to be ≥ 10◦
misorientation between the neighbouring grains. A small number of thinner
lines, particularly in the weld junction, represent misorientations ≥ 2◦ . The
coarse crystallographic grain size in the weld junction is apparent from the
large clusters of grains which have a similar colour–coding. As expected and
reported elsewhere [122, 123], it is different from the morphological grain
size observed using optical microscopy and ordinary scanning electron mi-
croscopy.
From the crystallographic data obtained, texture analysis was con-
ducted on different regions across the weld. Only results analysed near the
inner surface of the pipe are presented here for brevity. The numbers on all
the pole figures in this work indicate the pole density in “times random” unit.
Figure 4.19 shows a ferrite orientation image from a region 3 mm away from
the weld junction, i.e., with crystal orientations unaffected by the welding
process. Texture of the base metal is illustrated by the {100} pole figure.
However, when the pole figure is constructed for a region approximately
20 µm2 , it is clear that there the ferrite grains within this region have a large
range of orientations. Since one of the important factors influencing the
absorption of energy during cleavage fracture is the deflection of propagat-
ing cracks across grain boundaries, such a texture bodes well for toughness,
consistent with the base–plate data illustrated in Fig. 4.3.

57
58

Figure 4.16: Continuous ferrite orientation image of the as-welded sample, starting from 2 mm left (top left) to the middle of weld
junction (bottom right) near the outer surface of the X65HIC pipe.
59

Figure 4.17: Continuous ferrite orientation image of the as-welded sample, starting from 1 mm left (top left) to the middle of weld
junction (bottom right) at the mid-thickness of the X65HIC pipe.
60

Figure 4.18: Continuous ferrite orientation image of the as-welded sample, starting from 2 mm left (top left) to the middle of weld
junction (bottom right) near the inner surface of the X65HIC pipe.
 (a)

(b) (c)

Figure 4.19: (a) Orientation image of the base metal, 3 mm away from the
weld junction. (b) and (c) The {100} pole figures for the full area illustrated,
and for the boxed region in (a).

The situation changes dramatically in the affected regions of the weld.

The overall texture both in the region 1 mm away from the weld junction and
in the weld junction itself was obtained from the boxed regions in Fig. 4.18.
The size of the box is the same as the area in which the overall texture of
the base metal was gained, shown in Fig. 4.19a. The texture in these regions
is much sharper compared to that in base metal (cf. Fig. 4.19b and 4.20).
The coarse crystallographic grain size coincides vividly the poor toughness
recorded from Charpy tests [66] of the as–welded samples conducted on these
regions, Fig. 4.3a.
The orientation images from the post–weld heat-treated sample are
shown in Fig. 4.21, 4.22 and 4.23a. These images obtained from three dif-
ferent thicknesses of the pipe all start 1 mm away from the weld junction
at the top left and continues towards the middle of the weld junction at
the bottom right. They confirmed the previous observation that the post-
welding heat treatment unify the microstructure across the weld. The grain
size analysis based on the orientation images can also be done easily by the
HKL software. The grain is recognised by critical misorientation angle be-
tween neighbouring grains, which is set at 10◦ and by allowing to take the

61
Figure 4.20: {100} pole figures from the regions marked ‘i’ and ‘ii’ in Fig. 4.18,
showing the texture in a region 1 mm from the weld junction and in itself
near the inner surface of the X65HIC pipe.

closest route to complete the grain boundary. The aim of this analysis is
to measure the morphological grain size. The orientation images have the
closest similarity to the optical micrographs when the critical misorientation
angle is set at 10◦ . The amount of etching is critical for the metallographic
experiment, since the appearance of grain boundaries is very sensitive to the
etching by 2% nital. However the grain size obtained from the orientation
images is insensitive to the chosen misorientation angle, which will be dis-
cussed in the end of this section. Figure 4.24 shows results, in which each
line is constructed from an area with the same size of ‘i’ and ‘ii’ of Fig. 4.23.
It is clear that there is no significant difference of morphological grain size
in the region 1 mm away from the weld junction and in the weld junction
itself.
The overall texture in the corresponding regions 1 mm away from the
weld junction and in the weld junction itself was weakened relative to the
as-welded state (cf. Fig. 4.20 and Fig. 4.23b). Although at first glance
the refinement may appear to have happened, the texture analysis indicated
that the extent of the crystallographic refinement was limited. In the weld
junction, strong texture persists, as the typical result illustrated for region
(iii) in Fig. 4.23 indicates that the orientation differences between the grains
within the 20 µm interval regions are very small. Examination of the fracture
surface of a broken Charpy specimen revealed facets crossing many grains,
and of the same order of magnitude in scale to the clusters of similarly

62
oriented grains, Fig. 4.25. Hence, the post–weld heat treatment does not
result in an improvement of the low–temperature Charpy energy, Fig. 4.3b.
The {100} plane is the cleavage plane of the BCC iron crystal [86, 87];
so a high density of {100} planes parallel to rolling direction should lead to
a low fracture energy, referring to the specification of Charpy sample in this
study, Fig. 3.2b. It is useful therefore to examine the distribution of {100}
planes relative to the geometry of the Charpy test.
Fig. 4.26 shows the distribution of {100} poles relative to the trans-
verse direction, which is normal to the mean fracture plane of the Charpy
specimen. A high density of poles parallel to the transverse direction implies
a polycrystallography which is conducive to cleavage crack propagation. The
weld junction, both before and after the heat–treatment, retains the highest
density of cleavage planes in an unfavorable orientation, which is consis-
tent with the failure to improve toughness following reaustenitisation of the
welded region. The observed densities are much higher than for the region
3 mm away from the junction, i.e., the plate which is unaffected by the weld-
ing process. The region 1 mm from the junction has a generally low density
of {100} poles parallel to transverse direction, so the plot in Fig. 4.26 does
not explain why its toughness is poor in the as–welded condition, other than
due to the fact that its crystallographic grain size is large in the as–welded
condition, as is clear from region (i) in Figs. 4.18 and 4.20, where strong
texture exists.
Further analysis has been conducted in order to confirm the role of
crystallography in the cleavage fracture process. Strings of grains (defined
as those with misorientations across grain boundaries which are greater than
2◦ ) in the close proximity of the −40◦ C fracture at the weld junction were
analysed to examine approximately the continuity of {100} planes across
grains. For comparison purposes, a similar string from a region 3 mm from
the junction was also studied; the sample in this case was unbroken to avoid
the plastic deformation due to the high toughness of this area, Fig. 4.27.
The analysis is approximate because the edge–to–edge matching of cleavage
planes between adjacent grains is not considered, but rather is based on
the overall angle between planes from neighbouring crystals. The cleavage
crack propagates linearly along effective grains, and is deviated at boundaries
between effective grains, which was also pointed out in [124]. Fig. 4.28 shows
that the weld junction compares unfavourably relative to the unaffected base
plate since large cleavage facets are expected.

63
64

Figure 4.21: Continuous ferrite orientation image on the weld after post-welding heat treatment, starting from 1 mm left (top left)
to the middle of original weld junction (bottom right) near the outer surface of the X65HIC pipe.
65

Figure 4.22: Continuous ferrite orientation image on the weld after post-welding heat treatment, starting from 1 mm left (top left)
to the middle of original weld junction (bottom right) at the mid-thickness of the X65HIC pipe.
 (a)
66

(b)

Figure 4.23: (a)Continuous ferrite orientation image on the weld after post–weld heat treatment, starting from 1 mm left (top left)
to the middle of original weld junction (bottom right) near the inner surface of the X65HIC pipe. The region marked ‘iii’ is enclosed
in the small box. (b) {100} Pole figures of three corresponding regions in (a).
 (a)

(b)

(c)

Figure 4.24: (a) Grain size analysis of the post-welding heat-treated weld
from the X65HIC pipe. (a) Near the outer surface of the pipe. (b) At the
mid-thickness. (c) Near the inner surface.

67
Figure 4.25: Cross-sectional fracture surface near the V-notch of the Charpy
sample tested at −40◦ C with the V-notch located on the weld junction.

Figure 4.26: Plot of multiple of uniform density (M. U. D.) for the angle in
degrees, made by {100} poles to transverse direction. Data from Fig. 4.19b,
Fig. 4.20 and Fig. 4.23b (i and ii).

68
 c

Figure 4.27: Orientation image of grains for calculation of cleavage plane

propagation on (a) fractured sample after post-welding heat treatment, and
(b) base metal of as-welded sample. (c) Colour coding.

Figure 4.28: Calculated cleavage plane propagation curves.

69
 There is published work [65, 66] suggesting the effective grain size equal
to the average equivalent grain size obtained from the orientation image
with a specific misorientation angle. 12◦ was selected in [66] based on the
observation that single facets contain grains with up to 12◦ misorientation;
while 20◦ was chosen in [65], when the average equivalent grain size was
approaching the facet size. Similar analysis was carried out here on the
heat-treated weld junction near the inner surface of the pipe (region ‘ii’ of
Fig. 4.23). It is found that the equivalent grain diameter varies with the
misorientation angle as shown in Fig. 4.29. The average equivalent grain size
can not represent the facet size, which is up to ∼23 µm as shown in Fig. 4.25.
There is a large amount of small grains in the heat-treated weld junction
detected by the EBSD scan at a step size of 0.2 µm, but the total area fraction
of the grains no larger than the average grain size in the analysed region is
only around 0.1 as indicated by Fig. 4.24c. Therefore it is not appropriate
to use the average or any single equivalent grain size value to represent the
effective grain size of the microstructure without a normal distribution of
the grain size, like in the heat-treated weld. It is supported by the fact that
the facet size varies in the range approximately from 3 to 23 µm as shown in
Fig. 4.25.

Figure 4.29: The variance of equivalent grain size obtained from region ‘ii’
of Fig. 4.23 with the misorientation angle. Each line shows the maximum,
minimum and average values.

70
4.4 Conclusions

High frequency induction welded X65HIC steel

!
Mechanical properties
" $
" $
#
" %
$
Ductility Toughness

! !
Satisfactory Poor at the junction even after heat treatment
&
!
Possible causes

!
Abruptly microstructural change

No
!
Decarburisation

No
!
Remaining austenite

No
!
Inclusions

No
!
Crystallographic texture Yes

Figure 4.30: The route of investigation on X65HIC weld

The investigation on the weld of X65HIC pipe has gone through a jour-
ney as shown in Fig. 4.30. It appears that the poor toughness of the weld

71
junction of the induction weld is substantially related to firstly the crystallo-
graphically coarse grains present there after welding. Reaustenitisation that
occurs in the induction post-welding heat treatment does not improve the
situation at the weld junction, because the detrimental crystallographic char-
acteristics are reproduced on cooling. There is published work [97] indicating
the existence of an austenite memory effect during rapid reaustenisation ex-
periments. Based on this, studies on modified post-welding heat treatments
to improve the toughness in the weld–affected regions will be discussed in
Chapter 6.
Secondly the detrimental texture, i.e., orientation of the cleavage plane
prone to fracture is also found to be the cause of the low toughness of the
weld junction. Systematic analysis from this point of view is reported for the
welds from X65 grade of steels in Chapter 5.

72
Chapter 5

Welds from X65 Grade of

Steels

5.1 Mechanical properties

The mechanical properties of two kinds of pipes with different wall-thicknesses
at both as-welded and heat-treated states have been investigated in terms of
ductility and toughness.

5.1.1 Ductility
The results of cross-weld tensile tests are shown in Table 5.1. The as-welded
sample has the greatest ultimate strength but smallest elongation, and frac-
ture occurs in the base metal. Post-welding heat treatment decreases the ulti-
mate strength but increases the elongation of the as-welded sample. Fracture
happens in the weld junction. The final tensile strength is above the min-
imum required by the international standard, as discussed in section 2.2.2.

The post-welding heat treatment restored the ductility of the welds,

which is further proven by the bend tests carried out on weld segments of
the thick-walled X65 pipe. All of the 40 mm-wide sample, both as-welded and
heat-treated, fractured at the weld junction after face bends, but the heat-
treated sample demonstrated a higher degree of curvature prior to fracture
(cf. Fig. 5.1a and Fig. 5.1b). Three repeat tests were carried out in each
state.
During the side bend tests of the welds from the thick-walled pipe,
the as-welded sample fractured at the weld junction, as shown in Fig. 5.1c.
The heat-treated sample in comparison did not fracture after forming the
U shape as shown in Fig. 5.1d; however it reveals openings on the fusion

73
Table 5.1: Average and standard deviation of three tensile tests in each status
on high frequency induction welded X65 grade pipes.
(a) With 8.6 mm wall-thickness.
Ultimate tensile Fracture
strength / MPa elongation / %
Base metal 596±1 34.0±0
As-welded 621±1 9.2±5.3
Heat-treated 539±7 18.7±1.2

(b) With 14.8 mm wall-thickness.

Ultimate tensile Fracture
strength / MPa elongation / %
Base metal 626±1 33.9±1.3
As-welded 637±2 6.4±1.9
Heat-treated 573±8 24.9±5.2

(a) (b)

(c) (d)

Figure 5.1: Weld segments from the pipe with 14.8 mm wall-thickness after
bend testing. Scale indicated is in cm. (a) As-welded sample with 40 mm
width after face bend. (b) Heat-treated sample with 40 mm width after face
bend. (c) As-welded sample after side bend. (d) Heat-treated sample after
side bend.

surface, Fig. 5.2a. The heat-treated sample from the thin-walled pipe were
also tested and found to have developed openings on the fusion surface. The

74
RD–TD surface was observed by sectioning precisely on the location of one
opening. Several inclusions were discovered along the fusion surface below
the opening, as shown in Fig. 5.2b and c. EDX analysis showed that these
inclusions possess compositions consistent with manganese and silicon oxides,
Table 5.2.

Figure 5.2: Weld segments from the pipe with 14.8 mm wall-thickness after
side bend. (a) Openings on the fusion surface. (b) Microstructure of the
weld junction below the opening. (c) Inclusions and pull-outs.

(a)

(c)

(b)

Three 20 mm-wide heat-treated samples from both thick and thin-

walled pipes all passed the face bend test and they did not develop any
openings or cracks on the surface after deforming into a U shape, which is

75
Table 5.2: Compositions of the inclusions in Fig. 5.2c using EDX analysis.

Element Weight %
O 29.84±0.40
Al 1.19±0.10
Si 12.17±0.19
Ti 0.69±0.09
Mn 47.24±0.47
Fe 6.94±0.24
Nb 1.93±0.26

the best result obtained. In summary, due to the persistence of inclusions,

post-welding heat treatment did not restore the ductility of the welds to the
full extent, so some of the tested samples fractured and some others had
openings or cracks on the weld junction.

5.1.2 Toughness
Fig. 5.3 shows the results of cross-weld Charpy tests on both kinds of pipes.
As discussed in section 3.2.2, the experiments for the thin-walled pipe were
carried out on the samples with 23 of the size according to standard ISO 148-
1, while full size samples were used for the thick-walled pipe; so the absorbed
energies recorded from two kinds of pipes are not comparable. Each value on
the curves is an average of three measurements. The toughness of the welded
region improves as a consequence of the heat treatment when the V-notch is
located away from the weld junction, but at the location of the weld junction
the increase is not sufficient. Compared to the results from X65HIC pipe, the
thin-walled X65 pipe is in a similar situation. Again it is important to target
the cause for the increase of ductile brittle transition temperature at the weld
junction. The thick-walled X65 pipe demonstrated lower toughness at the
weld junction with no significant improvement at temperatures higher than
−40◦ C. Identifying and assessing the factors responsible for the low toughness
at the weld junctions of these two kinds of pipes needs sophisticated analysis,
which will be described in the next section.

5.2 Microstructural variance across the weld

Welds from two kinds of X65 pipes showed similar microstructure, so the
micrographs of the weld from only one kind of pipe are included in most of

76
 (a) X65 pipe with 8.6 mm wall-thickness.

(b) Scatter of the values at the weld junction

and the base metal from thin-walled X65 pipe.

(c) X65 pipe with 14.8 mm wall-thickness.

Figure 5.3: Charpy impact energy as a function of temperature.

the cases for brevity. More pictures are shown in Appendix A.

5.2.1 As-welded state

Referring to the observation on the welds from X65HIC pipe, the as-welded
sample can also be categorised into four zones: base metal, thermomechanically-

77
affected zone, heat-affected zone and weld junction. The microstructure of
the four zones was investigated. The base metal on the as-welded samples has
a microstructure of allotriomorphic ferrite plus pearlite, as shown in Fig. 5.4.
The larger fraction of pearlite compared to the X65HIC pipe is due to the
higher carbon concentration in the X65 grade of steels, Table 3.2.

Figure 5.4: Microstructure of base metal in the as-welded X65 pipe with 8.6
mm wall-thickness.

Different from the welds of X65HIC pipe, the grains from both the
thermomechanically-affected and heat-affected zones are aligned along the
directions of metal flow, which are different near the outer surface, at mid-
thickness and near the inner surface of the X65 pipe, as shown in Fig. 5.5.
The darker part of the images is within the heat-affected zone, and the rest
is in the thermomechanically affected zone. The weld junction is on the left
hand side of the region where the photos were taken and the base metal is on
the right hand side. The broader affected zone and greater influence from the
metal flow is due to the reduction in the welding speed from 22 m min−1 for
the X65HIC pipe to 15 and 12 m min−1 for the X65 pipes. This suggests that
there is a larger amount of heat input and deformation involved during the
forging of the two abutting edges when forming the X65 pipe compared to the

78
speeds removed for copyright reasons
 (a)

(b)

(c)

Figure 5.5: Optical micrographs of the thermomechanically-affected and

heat-affected zones on one side of the weld junction. (a) Near the outer
surface of the thin-walled X65 pipe. (b) At mid-thickness. (c) Near the
inner surface.

79
(a) (b)

(c) (d)

Figure 5.6: Microstructure of the thermomechanically-affected zone in the

as-welded sample from X65 pipe with 8.6 mm wall-thickness. (a) Optical
micrograph. (b) Scanning electron micrograph. Microstructure of the heat–
affected zone. (c) Optical micrograph. (d) Scanning electron micrograph.

Figure 5.7: Grain size analysis of the as-welded sample near the outer surface
of the X65 pipe with 8.6 mm wall-thickness. Each series represents the result
from an area with the size of 242.8×100 µm2 . The regions 1.84 and 2.53 mm
away from the middle of the weld junction belong to the heat-affected and
the thermomechanically-affected zones respectively.

80
situation with X65HIC pipe. This is expected from the greater concentration
of solutes in X65, and hence necessary to expel bad metal oxides or inclu-
sions which form during the heating and produce a sound weld. Both the
thermomechanically-affected and the heat-affected zones consist of allotri-
omorphic ferrite reformed from the grains which become austenite during
the welding process, and a second phase which appears dark in optical mi-
crographs and bright in scanning electron micrographs (cf. Fig. 5.6a and b, c
and d). This latter phase is most likely to be the so-called M–A constituent
and will be discussed later. The shape and size of the ferrite grains makes
the difference in microstructure between these two domains. It is anisotropic
in the thermomechanically-affected zone due to the severe pancaking of the
austenite before it transforms to ferrite [38, 116]. The heat-affected zone
in comparison has equiaxed and finer ferrite grains as a result of dynamic
recrystallisation due to more severe deformation, as it has closer proximity
to the weld junction. The grain size analysis (Fig. 5.7) was carried out on
orientation images using the critical misorientation angle of 10◦ as discussed
on page 62.
The microstructure at, and adjacent to the weld junction consists of a
mixture of allotriomorphic ferrite, Widmanstätten ferrite due to the coarser
austenite grains in this region, and M–A islands, Fig. 5.8. In the middle of the
weld junction, the islands are aligned along the normal direction (Fig. 5.8b)
throughout the wall-thickness except within 1 mm to both the inner and
outer surfaces. These strings of M–A islands are normally found along the
prior austenite grain boundaries [76, 125, 126] which on the fusion surface
do not have much mobility under impact from two abutting edges on both
sides. Results of nanoindentation, Fig. 5.9, showed that the strings are harder
than the surrounding phases in the matrix. The existence of the martensite
constituent was further confirmed by TEM. The TEM specimen was prepared
using focused ion beam milling, by which a slice thinner that 100 nm can be
cut just from a string, as shown in Fig. 5.10a and b. Figure 5.10c shows the
twinned martensite within the string, which is the characteristic morphology
of M–A islands in the high-strength low-alloy pipeline steels [127, 128].
Similar to the string of M–A islands found in the weld junction at the
mid-thickness of the X65HIC pipe, those in the X65 pipe are also manganese-
rich phases, as revealed by the elemental mapping using electron probe micro-
analysis in Fig. 5.11. There is possibly Mn segregation at the weld junction,
because Mn is fairly active in reaction with oxygen under high temperature

81
 (a)

(b)

Figure 5.8: Microstructure (a) adjacent to and (b) at the weld junction in
the as-welded pipe with 8.6 mm wall-thickness.

82
 (a)

(b)

Figure 5.9: (a) Nanoindentations on a string of M–A islands in the thin-

walled X65 as-welded sample. (b) Results of nanoindentation measurements.

83
 (a)

(b)

(c)

Figure 5.10: (a)Scanning electron micrograph of a string on the weld junction

of the as-welded sample from thin-walled X65 pipe. (b) A slice cut using
focused ion beam from the string. Micrograph is taken under the channel
contrast of focused ion beam. (c) Transmission electron micrograph of this
specimen.

84
 (a)

(b)

Figure 5.11: (a) Scanning electron micrograph of the weld junction of the
as-welded sample from thin-walled X65 pipe. (b) Mn mapping of (a) using
electron probe microanlaysis. This analysis can not be fully quantified due
to experimental limitation. Bigger number and warmer colour means higher
concentration of the element and vice versa.

85
at the joining surfaces during the welding process. The segregated Mn sta-
bilised the remaining austenite at the weld junction and formed the strings
of M–A islands. The Mn content in different phases during the equilibrium
transformation of thin X65 base material calculated by MTDATA [129] is
shown in Fig. 5.12. Manganese can remain in the austenite when cementite
does not form in the as-welded microstructure. The existence of remaining
austenite in the as-welded sample is further confirmed by X-ray diffraction,
Fig. 5.13. The quantity of austenite is small and is a part of the M–A con-
stituent so only a few of the austenite peaks could be identified in the X-ray
diffraction patterns. After tempering at 400 ◦ C for 2 h, the M–A islands
in the three zones of the as-welded sample all transformed into tempered
martensite, Fig. 5.14.
Weight percent of Mn in phase

22
20 Liquid
18 Ferrite
16 Austenite
Cementite
14
12
10
8
6
4
2
0
400 600 800 1000 1200 1400 1600 1800
Temperature /oC
(a)
Weight percent of Mn in phase

9
8
7
6
5
4
3
2
1
0
400 600 800 1000 1200 1400 1600 1800
Temperature /oC
(b)

Figure 5.12: Calculated equilibrium Mn content in different phases, as a

function of temperature for the thin X65 base material. (a) Allowing all the
possible phases to form. (b) With no cementite formed as in the as-welded
state.

86
(a)

(b)

Figure 5.13: X–ray diffraction patterns from weld segments of pipes with (a)
8.6 mm wall-thickness and (b) 14.8 mm wall-thickness.

87
 (a) Thermomechanically-affected zone.

(b) Heat–affected zone.

Figure 5.14: Microstructure of the thin-walled X65 as–welded sample after

tempering at 400 ◦ C for 2 h (continued on next page).

88
 (c) Weld junction.

Figure 5.14: Microstructure of the thin-walled X65 as–welded sample after

tempering at 400 ◦ C for 2 h.

5.2.2 After post-welding heat treatment

(a)

(b)

Figure 5.15: As-etched ND-TD surface of the weld from the pipe with 14.8
mm wall-thickness. (a) As-welded. (b) Heat-treated.

Similar to the X65HIC weld, post-welding heat treatment broadened

the affected zone on the welds from X65 pipes as shown in Fig. 5.15. The

89
microstructure across the weld also becomes uniform, and consists only of
allotriomorphic ferrite and pearlite, Fig. 5.16a and b. The fraction of the
pearlite decreases towards the unaffected steel, which makes the affected
area appear darker compared to the base metal on the as-etched surface,
Fig. 5.15b. X-ray diffraction patterns confirmed the absence of austenite
after post-welding heat treatment, as shown in Fig. 5.13a and b. Figure 5.16b
revealed clearly that the strings of M–A islands on the weld junction turned
into pearlite. The difference to the X65HIC heat-treated sample is that the
metal flow trace remained adjacent to the weld junction of the X65 heat-
treated samples (Fig. 5.17).

(a)

(b)

Figure 5.16: Scanning electron micrographs of the heat-treated X65 pipe

with 8.6 mm wall-thickness. (a) At the vicinity of the weld junction. (b) On
the weld junction.

90
 (a)

(b)

(c)

Figure 5.17: Optical micrographs showing the trace of the metal flow on one
side of the weld junction: (a) near the outer surface, (b) at mid-thickness
and (c) near the inner surface of the pipe.

5.3 Microhardness profile

The distributions of microhardness of the as–welded samples and heat–treated
samples are shown in Fig. 5.18a and c. The origin of the x-axis still corre-
91
sponds to the centre of the weld junction. On the as-welded samples, the
hardness showed higher values near the outer surface of the pipe due to the
surface hardening effect as discussed in section 4.2.1. There is an abrupt
drop of hardness on the weld junction of both pipes, which is the evidence
of decarburisation on the very location of the as-welded samples. The de-
carburised layer was observed in the normalised microstructure of the weld
junction using optical microscopy (Fig. 5.19).

(a) As-welded and heat-treated (solid (b) As-welded sample from X65 pipe
points) samples from X65 pipe with 8.6 with 8.6 mm wall-thickness after tem-
mm wall-thickness. pering.

(c) As-welded and heat-treated (solid (d) As-welded sample from X65 pipe
points) samples from X65 pipe with with 14.8 mm wall-thickness after tem-
14.8 mm wall-thickness. pering.

Figure 5.18: Microhardness profile across the welds from X65 pipes.

Tempering the as-welded samples at 400 ◦ C for 2 h slightly reduces

the hardness value across the weld (cf. Fig. 5.18a and b, c and d), which is
consistent with the existence of M–A islands in the as-welded microstructure.
Heat-treated samples showed a uniform distribution of hardness across
the weld (solid points in Fig. 5.18a and c), which corresponds well to the uni-
fied microstructure. There is no decrease of hardness on the weld junction,

92
 (a) ×50 magnification.

(b) ×200 magnification.

Figure 5.19: Optical micrograph of the welds from X65 pipe with 14.8 mm
wall-thickness after normalisation at 900 ◦ C for 30 min.

93
suggesting that post-welding heat treatment reduces the extent of decarburi-
sation.

5.4 Inclusions
The possibility of inclusions at the weld junction was investigated using ex-
tensive scanning electron microscopy together with EDX. Bearing in mind
that manganese and silicon oxides were found in the bent sample as discussed
in section 5.1.1, RD–TD surfaces with the weld junction in the middle were
cut and polished for investigation. In the heat-treated segment with the
size of 10 mm along the rolling direction from the pipe with 8.6 mm wall-
thickness, inclusions were revealed on the RD–TD surfaces near the outer
and inner surfaces and at the mid-thickness of the pipe as shown in Fig. 5.20.
Most of these inclusions aligned along the weld junction possess compositions
consistent with manganese and silicon oxides, as shown in Table 5.3. During
the welding process, these oxides formed by heating the edges of the steel
plate and did not get expelled by the forging. Post-welding heat treatment
was not able to attack the inclusions. At mid-thickness of the thin-walled X65
pipe, several aluminium and calcium rich oxides were also found (Fig. 5.20b
and Table 5.3). Fractography carried out on the tested Charpy sample from
the weld junction also revealed inclusions with similar compositions of man-
ganese and silicon oxides as shown in Fig. 5.21 and Table 5.3, which suggests
that the inclusions are detrimental to the toughness of the weld.

Table 5.3: Compositions for the inclusions (wt% with σ) by EDX.

O Al Si Ca Ti Mn Fe
Fig. 5.20a 31.43 2.04 16.46 0.59 1.48 45.07
2.92
±0.31 ±0.1 ±0.19 ±0.09 ±0.12 ±0.36
±0.27
Fig. 5.20b 31.50 2.79 14.98 0.62 1.40 40.50
8.22
±0.33 ±0.1 ±0.19 ±0.09 ±0.13 ±0.39
±0.31
Fig. 5.20b 34.77 27.71 1.29 11.69 – – 24.55
(circles) ±0.33 ±0.24 ±0.08 ±0.17 ±0.34
Fig. 5.20c 36.42 2.43 19.40 0.51 1.91 37.36 1.98
±0.35 ±0.11 ±0.22 ±0.08 ±0.13 ±0.38 ±0.25
Fig. 5.21 31.38 1.42 16.48 0.41 0.96 46.06 3.28
±0.26 ±0.08 ±0.13 ±0.05 ±0.07 ±0.24 ±0.15

Surprisingly when the same investigation on the RD–TD surfaces was

carried out on the heat-treated weld from the pipe with 14.8 mm wall-
thickness, no inclusions were found. Three pairs of Charpy fracture surfaces

94
 (a)

(b)

(c)

Figure 5.20: Scanning electron micrographs of the inclusions on the weld

junction of the heat-treated sample of X65 pipe with 8.6 mm wall-thickness
(a) Near the outer surface. (b) At mid-thickness. (c) Near the inner surface.

from the weld junction tested at −40◦ C were also examined. No clue of
inclusion initiated fracture was discovered. These three Charpy samples all
showed an absorbed energy value comparable to what the sample illustrated
in Fig. 4.14 gave. It might be the fact that the manganese and silicon oxides

95
Figure 5.21: Fracture surface at weld junction after post-weld heat treatment
from the pipe with 8.6 mm wall-thickness, tested at −40◦ C.

were not present continuously along the rolling direction, which is also the
welding direction, but it is more likely that there is another factor playing an
important role in causing the low toughness at the weld junction. In the case
of the pipe with 8.6 mm wall-thickness, the inclusions in Fig. 5.21 were found
on the tested sample giving the absorbed energy of 6 J, which is lower than
the mean value as shown in Fig. 5.3b. The highest value of absorbed energy
from the tests with the same condition is 44 J. This scatter is consistent
with the haphazard distribution of the inclusions along the weld. However,
it is not explained that even the highest value of absorbed energy at the
weld junction is still much lower than the result from the base metal at the
same testing temperature, which is above 120 J. The next section describes
the importance of crystallographic texture in affecting the toughness of the
weld.

5.5 Crystallographic analysis

Continuous scans on the ND–TD surface along the transverse direction near
the outer and inner surfaces, and at the mid-thickness of both pipes have
been carried out using the EBSD technique. Only the ferrite orientation im-
ages from the pipe with 14.8 mm wall thickness (Fig. 5.22–5.27) are displayed
here for brevity. More images are included in Appendix A. Austenite could
not be detected with confidence as discussed in section 4.3. Grain bound-
aries in the orientation images are designated mainly by the solid black lines
describing misorientation, φ ≥ 10◦ between the neighbouring grains and a
minority of thinner lines for φ ≥ 2◦ . The orientation images of the as-

96
welded sample showed clearly a gradual change of the microstructure from
base metal to thermomechanically-affected zone, heat-affected zone and to
the weld junction, Fig. 5.22–5.24. The post-welding heat treatment signif-
icantly unified the microstructure as shown in Fig. 5.25–5.27. What it did
not change completely is the texture originating from the welding.
Figure 5.28a shows a ferrite orientation image from the base metal,
which is an area 4 mm away from the weld junction on the as-welded sample.
The crystal orientations within this region are not uniformly distributed, as
illustrated by the {100} pole figure (Fig. 5.28c), but the ferrite grains have
a large range of orientations. More importantly, there is no accumulation
of {100} poles around the transverse direction. This means there are few
{100} planes parallel or nearly parallel to the ND–RD surface, which is the
fracture surface of the Charpy samples in this study, as shown in Fig. 3.2.
The {100} plane is the cleavage plane of BCC iron crystal [86, 87], so a low
density of {100} planes parallel to ND–RD surface should lead to a higher
fracture energy. Figure 5.28b shows the scattered grains having their {100}
planes oriented within 10◦ to the fracture surface, which represents only 3%
of the area in Fig. 5.28a. The results are consistent with the base-metal
toughness data illustrated in Fig. 5.3c. Conversely, a high density of {100}
planes parallel to the fracture surface leads to a lower toughness. The as-
welded junction is strongly textured with only one main orientation, and
there is high concentration of {100} poles around the transverse direction
(Fig. 5.28f). Some 19% of the area in Fig. 5.28d is occupied by clusters
of grains oriented for easy fracture, Fig. 5.28e, which accounts for the poor
toughness of this region, Fig. 5.3c. The intensity of the weld-junction texture
decreased following post-welding heat treatment, as illustrated in Fig. 5.28i,
but the accumulation of {100} planes around transverse direction was not
diminished. Bear in mind that the as-welded junction consists of both shear
products (Widmanstätten ferrite and M–A islands) and diffusional products
(allotriomorphic ferrite) from the transformation, so part of the prior austen-
ite grain structure formed during the welding process was reproduced by the
rapid heating of the induction heat treatment due to the austenite memory
effect [97]. Fig. 5.28h shows that grains possessing cleavage planes within
10◦ to the fracture surface still take up 10% of the area in Fig. 5.28g. This
explains why the improvement of toughness on the weld-junction by post-
welding heat treatment is not as much as being comparable to the base metal,
Fig. 5.3c.

97
 Table 5.4 gives a general view of the area percentage of {100} planes
similarly oriented with the fracture surface for both kinds of pipes. Each
value in the table is obtained from an EBSD scan of an area with the same
size of Fig. 5.28a, d or g, i.e., 242.8×100 µm2 . For both kinds of pipes,
the base metal has the lowest density of cleavage planes oriented for easy
fracture, while the as-welded junction has the highest. After post-welding
heat treatment, the density did not go down to the base metal level.
Similar to the calculation carried out to construct Fig. 4.27, three angles
between three {100} planes and the ND–RD surface of all the grains from
the scanned areas of base metal and heat-treated weld junction respectively
were calculated using the Euler angles obtained by EBSD technique. The
minimum of the three angles for each grain is selected to plot the distributions
shown in Fig. 5.29, with each line representing the same size of the areas
either in base metal or in weld junction. For both pipes, firstly there is higher
fraction of grains with unfavourably oriented cleavage planes in the heat-
treated weld junction compared to the base metal (Fig. 5.29a and c), which
makes the former more prone to cleavage. Secondly, base metal has higher
fraction of grains with their cleavage planes oriented in the angles between 40
and 50◦ to the fracture surface, which is the favourable orientation against
fracture. Besides, substantially larger number of grains contribute to this
range in the base metal than in the heat-treated weld junction (Fig. 5.29b
and d); since the former has finer grains, which is also noticeable in the
orientation images (cf. Fig. 5.28a and g). Larger number of grains oriented
against the fracture produce more deviation for the crack propagation across
more grain boundaries, which leads to higher absorbed energy and toughness.

5.6 Conclusions
It is found that the high frequency induction welded junctions from two X65
grade steel pipes with different wall-thickness possess unsatisfactory ductil-
ity and toughness after post-welding heat treatment, even though the mi-
crostructure across the weld was uniform and there is no decarburisation
and remaining austenite after post-welding heat treatment. Meticulous in-
vestigation has then identified three causes of the weakness: the existence
of non-metallic inclusions, crystallographic texture, and larger grain size.
Persistence of manganese and silicon rich oxides have a negative effect on
these properties, especially the ductility, but these inclusions only exist oc-

98
casionally along the welding direction. Low toughness in the weld-junction is
substantially related to the grain orientation favouring fracture. This ‘bad’
texture is produced during the welding process. Reaustenitisation during the
post-welding heat treatment reduces the intensity of the texture but dose not
alter it. The heat-treated weld-junction was also found to have coarser grains
than the base metal does, which helps to enlarge the difference in toughness
between the two samples.
The ∆ITT from base metal to the heat-treated weld junction of the
thick-walled X65 pipe is more than 80◦ C estimated from Fig. 5.3c. The
results of grain size analysis are shown in Fig. 5.30. They were carried out on
orientation images using the critical misorientation angle of 10◦ as discussed
on page 62. Taking the equivalent grain diameter value at 0.5 accumulative
area fraction from Fig. 5.30b, the contribution of the grain size difference to
∆ITT from base metal to the heat-treated weld junction of the thick-walled
X65 pipe is estimated at 35◦ C according to equation 2.14 by Pickering [69].
The contribution of crystallographic texture is assessed by correlating the
thick-walled X65 pipe to the X65HIC pipe. ∆ITT for the X65HIC pipe is
more than 40◦ C as estimated from Fig. 4.3, which is fully explained by the
crystallographic texture. The area percentage of grains containing {100}
planes within 10◦ to the ND–RD surface increased 7.6% from base metal to
the heat-treated weld junction of the X65HIC pipe. This increase is 5.7% in
the thick-walled X65 pipe as listed in Table 5.4b. Supposing the intensity of
the {100} texture has a linear relation with the transition temperature, the
change of crystallographic texture results in at least 30◦ C to ∆ITT from base
metal to the heat-treated weld junction of the thick-walled X65 pipe. The rest
of ∆ITT may be caused by the content of the non-metallic inclusions. This
assessment gives an attempt to predict ∆ITT after high frequency induction
welding by quantification of the potential contributors.

99
100

Figure 5.22: Ferrite orientation image of the as-welded sample near the outer surface of the thick-walled X65 pipe. The image is
continuous from the base metal at the top left to the middle of the weld junction at the bottom right.
101

Figure 5.23: Ferrite orientation image of the as-welded sample at mid-thickness of the thick-walled X65 pipe. The image is continuous
from the base metal at the top left to the middle of the weld junction at the bottom right.
102

Figure 5.24: Ferrite orientation image of the as-welded sample near the inner surface of the thick-walled X65 pipe. The image is
continuous from the base metal at the top left to the middle of the weld junction at the bottom right.
103

Figure 5.25: Continuous ferrite orientation image from 1 mm left towards the middle of the weld junction on the sample after
post-welding heat treatment near the outer surface of the thick-walled X65 pipe.
104

Figure 5.26: Continuous ferrite orientation image from 1 mm left towards the middle of the weld junction on the sample after
post-welding heat treatment at mid-thickness of the thick-walled X65 pipe.
105

Figure 5.27: Continuous ferrite orientation image from 1 mm left towards the middle of the weld junction on the sample after
post-welding heat treatment near the inner surface of the thick-walled X65 pipe.
 (a) (b) (c)
106

(d) (e) (f)

(g) (h) (i)

Figure 5.28: (a) Orientation image of base metal near the outer surface of the pipe with 14.8 mm wall-thickness. (b) Orientation
image of grains in the same area of (a), oriented in the way that their {100} planes are within 10◦ to ND–RD surface. (c) Pole figure
from the area of (a). (d)(e)(f) and (g)(h)(i) Equivalent images for the as-welded and the heat-treated weld junctions respectively.
Table 5.4: Area percentage of grains containing {100} planes within 10◦ to
the ND–RD surface.
(a) Pipe with 8.6 mm wall-thickness.
Weld junction
Base metal
As-welded Heat-treated
Near outer surface 2 12 9
At mid-thickness 2 24 10
Near inner surface 2 12 7
Average 2 16 8.7

(b) Pipe with 14.8 mm wall-thickness.

Weld junction
Base metal
As-welded Heat-treated
Near outer surface 3 19 10
At mid-thickness 3 12 10
Near inner surface 4 14 7
Average 3.3 15 9

107
 50 6000

Percentage of grains / %
40 5000

Number of grains
4000
30
3000
20
2000
10 1000

0 0
0-

10

20

30

40

50

60

70

80

0-

10

20

30

40

50

60

70

80
10

10
-2

-3

-4

-5

-6

-7

-8

-9

-2

-3

-4

-5

-6

-7

-8

-9
0

0
o o
Angle / Angle /

(a) Pipe with 8.6 mm wall-thickness. (b) Pipe with 8.6 mm wall-thickness.
108

50 4000
Base metal
3500 Heat-treated weld junction
Percentage of grains / %

40
3000

Number of grains
30 2500
2000
20 1500
1000
10
500
0 0
0-

10

20

30

40

50

60

70

80

0-

10

20

30

40

50

60

70

80
10

10
-2

-3

-4

-5

-6

-7

-8

-9

-2

-3

-4

-5

-6

-7

-8

-9
0

0
o o
Angle / Angle /

(c) Pipe with 14.8 mm wall-thickness. (d) Pipe with 14.8 mm wall-thickness.

Figure 5.29: Distribution of the minimum angle between {100} plane and the ND–RD surface for all the grains scanned by EBSD
technique.
 (a) Pipe with 8.6 mm wall-thickness.

(b) Pipe with 14.8 mm wall-thickness.

Figure 5.30: Grain size analysis carried out on the areas with the same size
on the base metal and heat-treated weld junction.

109
Chapter 6

Simulations of the High

Frequency Induction Welding
and Post-Welding Heat
Treatment

6.1 Welds from X65HIC pipe

6.1.1 Welding process
Experiments were carried out to simulate the induction welding process on
X65HIC base metal, the thermal history of which is shown in Fig. 6.1. The
sample was heated up at 20 ◦ C s−1 to 1450 ◦ C and then cooled down at the
same rate to room temperature. The peak temperature is chosen because the
pyrometer reading for the temperature of the pipe just after passing through
the induction coil is in a range of 1400–1500 ◦ C and the sample should not be
melted due to an experimental limitation of the equipment used. Calculations
using MTDATA [129] showed that the X65HIC steel should not melt until
1500 ◦ C under equilibrium conditions, Fig. 6.2. This calculation takes into
account 11 elements and 4 phases, which are Fe, C, Mn, P, Si, Al, Cu, Ni,
Cr, Nb, V, ferrite, austenite, cementite and liquid. The peak temperature
of 1450 ◦ C should match that of the actual temperature experienced by the
material in close vicinity of a weld junction, since any molten material is
expelled due to the forging pressure.
The microstructure obtained by the simulation is shown in Fig. 6.3. It
consists of platelets of ferrite, which has a great similarity to the microstruc-
ture of granular bainitic ferrite and acicular ferrite which have been referred
in a large amount of literature as typical microstructure in continuously-
cooled low-carbon steels [130–133]. The former is also known as granular

110
 Figure 6.1: Simulation of induction welding on X65HIC base metal.

1
Liquid
Ferrite
Austenite
0.8 Cementite
Mass fraction

0.6

0.4

0.2

0
600 700 800 900 1000 1100 1200 1300 1400 1500 1600
Temperature / oC

Figure 6.2: Calculated equilibrium transformation of X65HIC base metal.

111
bainite [134]. The black regions in the optical micrograph, Fig. 6.3, are most
likely to be dispersed retained austenite or M–A islands in an either equiaxed
or elongated morphology. Confirmation of these phases needs further charac-
terisation, which was not pursued in this study because this microstructure
is not similar to that of any zone observed in the as–welded sample (sec-
tion 4.2.1). This suggests that the cooling rate after the welding process
may not be as high as the one used in the simulation. The induction heating
however should be much faster, but there are no data available for the heat-
ing rate associated with induction in the production line. Additionally, the
welding process in practice involves force to press the two abutting edges of
the steel plate together, which is not applied in this simulation.

Figure 6.3: Microstructure of X65HIC base metal after simulated welding

process.

6.1.2 First cycle of post-welding heat treatment

Since production samples extracted after the first cycle of the post-welding
heat treatment were not available, simulation of the heat treatment (austeni-
sation followed by cooling) was carried out on the as-welded sample in order
to understand every step of the whole process. After the austenisation, air
cooling gave the microstructure at the weld junction as shown in Fig. 6.4,
consisting of ferrite and pearlite. At the mid-thickness of the pipe, there is
pearlite aligned in a line (Fig. 6.4b), which is supposed to transform from

112
the string in the as-welded sample (page 49). This demonstrates that accel-
erated cooling is necessary after austenistation to produce a finer final grain
structure content with optimum mechanical properties.
Austenisation followed by quenching in iced brine gives the microstruc-
ture shown in Fig. 6.5. It should be similar to that from the industrial
process, although the heating rate by induction (estimated ∼ 50 ◦ C s−1 )
should be more rapid than by placing the sample into a furnace at 960 ◦ C
(∼ 10 ◦ C s−1 ). The microstructure at the weld junction, especially near the
outer (Fig. 6.5a) and inner (Fig. 6.5e) surface of the pipe, consists of grain
boundary allotriomorphs of ferrite and packets of plate-like grains which are
likely to be low-carbon martensite or bainite. The microstructure at the
mid-thickness of the pipe (Fig. 6.5c) mainly consists of fine-grained allotri-
omorphic ferrite. This difference is probably due to the faster cooling rate
near the edge of the sample during quenching. The microstructure at the
domain of original base metal (Fig. 6.5b, d and f) is similar to the one at
the weld junction except for a finer grain size, which is most likely caused by
the smaller austenite grains formed in the base metal. Since coarse austenite
grains form at the weld junction (page 49), the austenisation during post-
welding heat treatment does not reduce the autenite grain size in the weld
junction to that in the base metal due to the austenite memory effect [97].

6.1.3 Modified post-welding heat treatments

In the industrial process, the peak temperature for the first cycle post-welding
heat treatment is 960 ◦ C, substantially above the calculated Ae3 temperature
of the steel, 870 ◦ C as shown in Fig. 6.2. The material is supposed to be fully
austenised during the heat treatment. As discussed on page 30, the prior
austenite grain is reconstructed above Ac3 due to the austenite memory ef-
fect, so an intercritical annealing followed by quenching and then tempering
is proposed for the post-welding heat treatment in order to prevent the aci-
cular austenite formed above Ac1 coalescing, which may lead to the grain
refinement as evidenced in [98].

Temperature removed for copyright reasons

113
 (a)

(b)

(c)

Figure 6.4: Microstructure at the weld junction after heating the X65HIC
as-welded sample to 960 ◦ C followed by an air cool. (a) Near the outer surface
of the pipe. (b) At the mid-thickness (c) Near the inner surface.

114
115

(a) Optical microgragh taken at the location of original weld junc- (b) Optical micrograph taken at the location of original base metal.
tion.

Figure 6.5: Microstructure of as–welded sample after heating to 960 ◦ C followed by quenching in iced brine near the outer surface of
the X65HIC pipe (continued on next page).
116

(c) Optical microgragh taken at the location of original weld junc- (d) Optical micrograph taken at the location of original base metal.
tion.

Figure 6.5: Microstructure of as–welded sample after heating to 960 ◦ C followed by quenching in iced brine at the mid-thickness of
the X65HIC pipe (continued on next page).
117

(e) Optical microgragh taken at the location of original weld junc- (f) Optical micrograph taken at the location of original base metal.
tion.

Figure 6.5: Microstructure of as–welded sample after heating to 960 ◦ C followed by quenching in iced brine near the inner surface of
the X65HIC pipe.
 The calculated Ae1 and Ae3 of X65HIC are 700 and 870 ◦ C respectively
(Fig. 6.2). The Ac1 is 838±13 ◦ C and the Ac3 is 911±2 ◦ C when the sam-
ple is heated at 20 ◦ C s−1 using the thermomechanical simulator, as shown
in Fig. 6.1. The heating rate applied in the experiments by placing sample
into the furnace with preset temperatures was around 10 ◦ C s−1 , which sug-
gests the Ac1 to Ac3 should be between 700–870 ◦ C and 838–911 ◦ C; since
the austenisation temperature increases with the heating rate. Therefore the
peak temperature for the first intercritical annealing heat treatment (Treat-
ment 1) was chosen to be 830 ◦ C. The temperature profile of this experiment
is shown in Fig. 6.6. The second cycle of heat treatment is a simulation of
that in the industrial process, which is tempering the weld specimen up to
480 ◦ C followed by air cooling. The microstructure of the weld (Fig. 6.7) did
not change much following heat treatment. It can still be categorised into
four domains similar to the as-welded sample (Fig. 4.5a and 4.6). This indi-
cates that the intercritical annealing treatment at 830 ◦ C was not effective.
The Ac1 to Ac3 range of this material at the heating rate used is approaching
838–911 ◦ C obtained at the heating rate of 20 ◦ C s−1 , so 830 ◦ C is probably
too low.

Figure 6.6: Thermal history of modified post-welding heat treatment ‘Treat-

ment 1’ on X65HIC as–welded sample.

118
 (a) Base metal. (b) Thermomechanically affected zone.

(c) Heat-affected zone. (d) Weld junction.

Figure 6.7: Microstructure of the X65HIC as–welded sample after ‘Treat-

ment 1’.

119
 A second intercritical annealing heat treatment (Treatment 2) was con-
ducted as shown in Fig. 6.8, with the peak temperature of the first cycle
raised to 890 ◦ C and the second cycle is still a simulation of the industrial
process. The resulted microstructure becomes more uniform with no notice-
able difference as a function of distance from the weld junction as shown in
Fig. 6.9a–c, although a relatively coarse microstructure still persists in the
weld junction.

Figure 6.8: Thermal history of ‘Treatment 2’ on X65HIC as–welded sample.

Orientation images revealed a large crystallographic grain size at the

weld junction, inherited from the as-welded sample. According to Fig. 6.10,
the industrially produced sample still has a relatively more uniform distri-
bution of crystal orientations. It is possibly caused by high intercritical
annealing temperature, at which the acicular austenite has already started
to coalesce into coarse austenite grains. The industrial process on the other
hand, by austenising above Ac3 , possibly induced an extent of recrystalli-
sation, which led to grain refinement. Supposing the newly formed acicular
austenite did not coalesce at 890 ◦ C, as parts of the prior austenite grain they
may transform upon cooling into the martensite or bainite with similar ori-
entations to the ones after welding. Thus the heat-treated weld maintained
the coarse crystallographic grains although the morphological grain size was
reduced.

120
 (a) At the location originally belonging (b) Thermomechanically affected zone.
to the base metal.

(c) Heat-affected zone. (d) Weld junction.

Figure 6.9: Microstructure of the X65HIC as–welded sample after ‘Treat-

ment 2’.

121
 (a)
122

(b)

(c)

Figure 6.10: Ferrite orientation images in the vicinity of weld junction near the outer surface of the X65HIC pipe. (a) As-welded
sample. (b) After ‘Treatment 2’ (c) After post-welding heat treatment.
 The modified post-welding heat treatment was then aimed at austeni-
sation at higher temperature than 960 ◦ C which is employed during the in-
dustrial process, in order to promote the austenite recrystallisation and grain
refinement. The new process, ‘Treatment 3’, involves heating the as-welded
sample up to 1000 ◦ C followed by quenching in water and then tempering up
to 480 ◦ C finished by air cooling to ambient temperature, Fig. 6.11.

Figure 6.11: Thermal history of ‘Treatment 3’ on X65HIC as–welded sample.

After ‘Treatment 3’, the microstructure is clearly uniform across the

weld. Fig. 6.12 shows the scanning electron micrographs obtained at the
locations of the original weld junction and base metal. The microstructure
contains fine plates of ferrite. The bright regions in the scanning electron
micrographs are most likely to be fine carbides formed during the tempering.
Compared to the industrially heat-treated weld junction, the sample
after ‘Treatment 3’ demonstrates a more uniform distribution of the crystal
orientations as shown in Fig. 6.13. Pole figures generated from these data
further prove this point, since the texture in the ‘Treatment 3’ sample is
weaker than in the industrial sample (cf. Fig. 6.14abc and def). Indeed
there is relatively little accumulation of cleavage planes ({100}) parallel to
the fracture surface (normal to TD) in the sample after ‘Treatment 3’. Grains
possesing {100} planes oriented within 10◦ to the fracture plane occupy only
6% of the scanned area compared to 10.3% in the industrial sample, Fig. 6.15.

123
124

(a) Scanning electron microgragh taken at the location of original weld (b) Scanning electron microgragh taken at the location of original base
junction. metal.

Figure 6.12: Microstructure of the as–welded sample after ‘Treatment 3’ near the outer surface of the X65HIC pipe (continued on
next page).
125

(c) Scanning electron microgragh taken at the location of original weld (d) Scanning electron microgragh taken at the location of original base
junction. metal.

Figure 6.12: Microstructure of the as–welded sample after ‘Treatment 3’ at mid-thickness of the X65HIC pipe (continued on next
page).
126

(e) Scanning electron microgragh taken at the location of original weld (f) Scanning electron microgragh taken at the location of original base
junction. metal.

Figure 6.12: Microstructure of the as–welded sample after ‘Treatment 3’ near the inner surface of the X65HIC pipe.
 (a) (d)
127

(b) (e)

(c) (f)

Figure 6.13: Ferrite orientation images of the weld junctions on the sample after ‘Treatment 3’ (a) near the outer surface, (b) at the
mid-thickness and (c) near the inner surface of the X65HIC pipe; and on the sample after industrial post-welding heat treatment
(d) near outer surface, (e) at mid-thickness and (f) near inner surface.
(a) (d)

(b) (e)

(c) (f)

Figure 6.14: {100} pole figures obtained from corresponding orientation im-
ages in Fig. 6.13.

128
 (a) (d)
129

(b) (e)

(c) (f)

Figure 6.15: Grains owning {100} planes oriented within 10◦ to the ND–RD surface in the corresponding orientation images of
Fig. 6.13.
 All of this evidence supports the notion that ‘Treatment 3’ reduces
the fracture-prone crystallographic texture on the weld junction which forms
during welding process, and persists after the industrial post-welding heat
treatment. Figures 6.13 and 6.15 also show that the morphological grain
size of ‘Treatment 3’ sample is refined compared to the industrial sample
as confirmed by the grain size analysis shown in Fig. 6.16. Finer grains are
essential to improved toughness.

Figure 6.16: Grain size analysis based on the orientation images in Fig. 6.13
from Modified3 sample and industrially heat-treated sample respectively.

6.2 Welds from thin-walled X65 pipe

Heat treatments were carried out on as-welded samples from thin-walled X65
pipe in order to assess the relationship with toughness of the weld. The first
comparison was made between single- and double-cycle post-welding heat
treatments. Three pairs of simulations with and without the second cycle
were carried out under the conditions shown in Table 6.1. All the samples
were quenched at the end of the first cycle of the heat treatment but air
cooled at the end of the second. The toughness of the simulated samples was
measured with the Charpy V-notch located on the weld junction. Figure 6.17
shows that the second cycle of the post-welding heat treatment is absolutely
essential to improve the toughness of the weld. After the first cycle of the
heat treatment, the sample shows a microstructure (Fig. 6.18) similar to the
X65HIC as-welded sample after its first cycle heat treatment (Fig. 6.5a and
e), containing allotriomorphic ferrite and low-carbon martensite which may

130
not be a desirable phase for satisfactory toughness. The hardness of the
sample remains at the level of as-welded sample (∼260 HV), much higher
than the industrially heat-treated sample (∼200 HV) as shown in Fig. 5.18a.

Table 6.1: Simulated post-welding heat treatments on as-welded samples

from X65 pipe with 8.6 mm wall thickness. ‘-C’ represents complete post-
welding heat treatments and ‘-Q’ represents heat treatments stopped after
the first cycle.

1st cycle 2nd cycle

Sample Peak Holding Peak Holding
temperature / C time / s temperature / C time / s
◦ ◦

A-C 1054 12 922 12

A-Q 1054 12 – –
B-C 1054 12 890 23
C-C 1054 12 922 23
D-C 1054 12 950 23
E-C 1150 12 922 23
E-Q 1150 12 – –
F-C 1054 23 922 23
G-C 1054 27 922 38.4
G-Q 1054 27 – –

Figure 6.17: Difference in toughness of the simulated samples with double-

or single-cycle heat treatments. The error bar represents two standard devi-
ations of the Charpy impact tests results.

The next stage was to investigate a set of simulations with variation

in the parameters of the first cycle of the post-welding heat treatment and

131
Figure 6.18: Optical micrograph of the weld junction on the simulated sample
‘A-Q’. (This figure is taken at the ArcelorMittal Research and Development
laboratory in Ghent.)

the same conditions for the second cycle (holding at the peak temperature of
922 ◦ C for 23 s) as shown in Table 6.1. Figure 6.19 shows that the toughness
of three samples does not vary significantly. Similarly, the samples after the
same first cycle heat treatment, holding at the peak temperature of 1054 ◦ C
for 12 s as shown in Table 6.1, the toughness did not vary significantly with
the different parameters of the second cycle, Fig. 6.20.
Charpy impact tests over a range of temperatures were carried out on
sample ‘C-C’, whose thermal history is shown in Fig. 6.21, with impact tough-
ness data compared with the industrially heat-treated sample in Fig. 6.22.
The weld junction of sample ‘C-C’ demonstrated significantly higher ab-
sorbed energy in the low temperature range than the industrially heat-treated
junction. The ultimate tensile strength is comparable for the two samples,
although the hardness of the ‘C-C’ sample is slightly lower (Fig. 6.23).
Microstructure of the sample ‘C-C’ contains ferrite as the matrix and
minor quantities of pearlite and austenite, which are shown in white, black
and grey in the optical micrograph, Fig. 6.24a. The scanning electron micro-

132
Figure 6.19: Variation in toughness of the simulated samples with different
parameters of the first cycle of the heat treatment. The error bar represents
two standard deviations of the Charpy impact tests results.

Figure 6.20: Variation in toughness of the simulated samples with differ-

ent parameters of the second cycle of the heat treatment. The error bar
represents two standard deviations of the Charpy impact tests results.

graph, Fig. 6.24b, clearly revealed that the pearlite started to form from some
remaining austenite grains (white featureless). The existence of austenite is
further confirmed by the X-ray diffraction pattern as shown in Fig. 6.25. It
suggests that the cooling rate at the end of industrial post-welding heat treat-
ment is slower than the one applied on sample ‘C-C’, because the industrially
heat-treated sample does not contain any retain austenite, as confirmed in
Fig. 5.13a. The austenite had enough time to transform into pearlite. This

133
Figure 6.21: Thermal history of the simulated post-welding heat treatment
on sample ‘C-C’.

Figure 6.22: Results of Charpy impact tests on simulated sample ‘C-C’ and
the industrially heat-treated sample from the X65 pipe with 8.6 mm wall-
thickness.

trace amount of austenite in sample ‘C-C’ seems not to be detrimental to

the toughness of the weld junction.
Crystallographic analysis using the EBSD technique was also carried

134
Figure 6.23: Results of cross-weld tensile and microhardness tests on the
simulated sample ‘C-C’ and the industrially heat-treated sample from the
X65 pipe with 8.6 mm wall-thickness. Three tensile tests were carried out
and the value of hardness was obtained from 10 indents on each sample.

(a) Optical micrograph.

Figure 6.24: Microstructure of the weld junction on sample ‘C-C’. ‘P’ and ‘γ’
represent the examples of pearlite and austenite (continued on next page).

135
 (b) Scanning electron micrograph.

Figure 6.24: Microstructure of the weld junction on sample ‘C-C’. ‘P’ and
‘γ’ represent the examples of pearlite and austenite.

Figure 6.25: X-ray diffraction pattern obtained on the simulated sample ‘C-
C’.

136
(a) (d)

(b) (e)

(c) (f)

Figure 6.26: Ferrite {100} pole figures generated from scanned areas by
EBSD technique with the same size on the weld junctions of sample ‘C-C’
(a) near the outer surface, (b) at the mid-thickness and (c) near the inner
surface of the X65 pipe with 8.6 mm wall thickness; and on the sample
after industrial post-welding heat treatment (d) near outer surface, (e) at
mid-thickness and (f) near inner surface.

137
out on the sample ‘C-C’. Comparison of the ferrite {100} pole figures ob-
tained from the areas with same size on the weld junctions of both sample
‘C-C’ and the industrially heat-treated sample at three pipe thicknesses is
shown in Fig. 6.26. The crystal orientations are more uniformly distributed
on the weld junction of ‘C-C’ sample and there is obviously a lower density of
{100} poles normal to the ND–RD fracture surface compared to the industri-
ally heat-treated sample. Grains possessing {100} planes oriented within 10◦
to the ND–RD surface occupy 5% of the scanned area on the weld junction
of the ‘C-C’ sample; in comparison this number rises to 8.7% of the scanned
area with same size on the weld junction of the sample after industrial heat
treatment. These results are consistent with the higher toughness of the ‘C-
C’ weld junction than the industrially heat-treated one. Furthermore, the
weld junction of the former sample has finer grains as shown in Fig. 6.27,
which leads to better toughness.

Figure 6.27: Grain size analysis carried out on the areas with the same size
on the weld junctions of the simulated sample ‘C-C’ and industrially heat-
treated sample from the thin-walled X65 pipe.

6.3 Conclusions
For the welds from X65HIC grade of steel, an approximate simulation of the
high frequency induction welding has been carried out using a thermome-
chanical simulator. Although a microstructure identical to the commercial
weld was not obtained, the experiment suggests that the cooling rate after
welding in the production line might be lower than the one used in the sim-
ulation, which is 20 ◦ C s−1 ; because bainitic ferrite, if it is confirmed in the

138
simulated microstructure, results from faster cooling than that which gives
Widmanstätten ferrite in the as-welded sample.
Simulations of the first cycle of the post-welding heat treatment were
carried out on both X65HIC and thin-walled X65 welds in a chamber fur-
nace and a Gleeble simulator respectively. Both simulated samples showed a
microstructure consisting of allotriomorphic ferrite and a plate-like structure
which is most likely to be low-carbon martensite or bainite, and is not a
desirable microstructure for satisfactory toughness property of the weld; so
the second cycle of the heat treatment was proven to be useful in improving
toughness.
Modified post-welding heat treatment could not give a sufficient inter-
critical annealing by setting the peak temperature of first cycle at 830 ◦ C,
since the Ac1 was raised well above Ae1 by placing the as-welded sample into
the furnace with the preset temperature. Intercritical annealing at 890 ◦ C
and tempering on the weld from X65HIC pipe did not remove crystallo-
graphically coarse grains. The lack of method to predict the temperature at
which the acicular austenite starts to coalesce and the shortage of as-welded
sample laid big obstacles to pursue more intercritical annealing experiments.
Austenisation at a higher temperature than in the industrial process, fol-
lowed by similar quenching and tempering, resulted in a much better struc-
ture and crystallography. This may be because the austenite memory effect
was removed. The resulting crystallographic orientations are more uniformly
distributed and should make fracture more difficult.
Variation in the parameters (peak temperature and holding time for
both cycles) of the post-welding heat treatment in a range, did not signif-
icantly affect the upper shelf energy of the weld junction from the thin-
walled X65 pipe. However one selected sample that was cooled possibly
faster at the end of the simulation than the industrially heat-treated sam-
ple, achieved higher lower shelf energy at the weld junction. It is evident
that uniformly distributed crystal orientations and smaller grain sizes lead
to improved toughness.

139
Chapter 7

General Conclusions and

Proposed Future Research

The work presented in this thesis had the primary goal of resolving an obser-
vation that the toughness at the induction welded junction is poor in spite
of sophisticated heat treatments. Two varieties of X65 pipeline steel have
been studied, one designed to resist hydrogen induced cracking. After much
investigation of factors associated with inclusions or decarburisation, it was
discovered that the problem is associated with the formation of pernicious
coarse “crystallographic grains” which do not adequately resist fracture. It is
believed that this texture, which manifests itself as the clustering of similarly
oriented grains, is produced during the welding process and is regenerated
upon subsequent heat treatment involving rapid heating and cooling due to
an austenite memory effect which has previously been reported in the pub-
lished literature.
Attempts were made therefore to modify the post-welding heat treat-
ment aimed at the removal of this ‘memory’. Simulation using austenisation
at higher temperature than the industrial process succeeded in producing a
weld junction containing a more uniformly distributed set of crystallographic
orientations and at the same time, finer crystallographic and morphological
grain sizes. This new treatment can in principle be adopted in production,
but steel availability restricted validation in the full scale production facility.
This constitutes future work in the event that X65HIC pipe is ordered by
customers. However such a trial should be preceded by detailed mechanical
characterisation on a laboratory scale. The improvement of toughness on the
weld junction through a simple change of the processing parameter has the
potential to enormously affect the pipeline-making industry.
When it comes to the normal X65 pipes which are less pure, manufac-

140
tured by the same technique, inclusions become an issue. The weld junctions
of two pipes with different wall-thicknesses were found to develop openings
at the locations of embedded inclusions during bend tests. Nevertheless,
the inclusions are infrequently located at the positions of poorest toughness
and hence are deemed to be of secondary consequence. Similar to the welds
from X65HIC pipe, the toughness of the weld junction after post-welding
heat treatment is significantly lower than that of the base metal. Crystal-
lographic analysis showed cleavage-prone texture persistent throughout the
process, which adversely affects toughness, more so than the inclusions which
are occasionally present along the rolling direction of the pipe.
The frequency of inclusions can be reduced by expelling more material
from the weld junction during the induction welding process. Most of the
inclusions are manganese and silicon oxides which form due to oxidation on
the surfaces of the abutting edges of the steel plate during the welding pro-
cess. Employing more force onto the pressure rolls or a redesign of the layout
in the welding mill is recommended to eject the material uniformly along the
rolling/welding direction. Both methods need systematic calculations using
actual data obtained on the production line.
More carefully designed post-welding heat treatment should be able to
modify the crystallographic texture in the weld junction. The present work
has shed light on how to achieve this. More studies are proposed to identify
any difference between the industrial process and the laboratory simulations,
e.g., the effect of cooling rate at the end of heat treatment on the toughness
of the weld junction.
In practice, the X65HIC pipe is heat-treated using an austenisation
and then a tempering cycle. The X65 pipes on the other hand are given
double austenisation cycles. The initial trial of austenisation followed by
tempering at 600◦ C on the thin-walled X65 pipe showed a decrease in tough-
ness compared to the one that had been austenised twice, but the reason is
not understood. Further research should be carried out on the post-welding
heat treatment involving austenisation plus tempering by either experimen-
tal or computational simulations, because the X65HIC pipe following this
route gave satisfactory mechanical properties and it will build up on the un-
derstanding of the “heat treatment—microstucture—mechanical properties”
relation. Assuming that double austenisation does give better toughness, the
weld from X65HIC pipe should be tested using this approach in the hope of
enhancing toughness.

141
temperature removed for copyright reasons
 Post-welding heat treatment using high frequency induction is a com-
plex process involving numerous variables, e.g., the chemical compositions
of the material, detailed thermal treatment and parameters associated with
the induction process. It is feasible that appropriate industrial data can be
accumulated in the long term to enable them to be subjected to a Bayesian
neural network analysis [135, 136] in order to recognise patterns within these
complex phenomenon. If such data are not likely to become available, they
should be collected through simulations such as those described in this thesis.
The current work seems to have made a good start in this respect.

142
Appendix A

Additional supporting evidence

for Welds from X65 Pipes

(a) Base metal.

Figure A.1: Microstructure of the weld from the as-welded X65 pipe with
14.8 mm wall-thickness (continued on next page).

143
 (b) Thermomechanically-affected zone.

(c) Heat-affected zone.

Figure A.1: Microstructure of the weld from the as-welded X65 pipe with
14.8 mm wall-thickness (continued on next page).

144
 (d) Weld junction.

Figure A.1: Microstructure of the weld from the as-welded X65 pipe with
14.8 mm wall-thickness.

(a) Adjacent to the weld junction.

Figure A.2: Microstructure of the heat-treated weld from X65 pipe with 14.8
mm wall-thickness (continued on next page).

145
 (b) On the weld junction.

Figure A.2: Microstructure of the heat-treated weld from X65 pipe with 14.8
mm wall-thickness.

146
147

(a) Near the outer surface.

Figure A.3: Ferrite orientation image of the as-welded sample from X65 pipe with 8.6 mm wall-thickness. The image is continuous
from the base metal at the top left to the middle of the weld junction at the bottom right (continued on next page).
148

(b) At mid-thickness.

Figure A.3: Ferrite orientation image of the as-welded sample from X65 pipe with 8.6 mm wall-thickness. The image is continuous
from the base metal at the top left to the middle of the weld junction at the bottom right (continued on next page).
 (c) Near inner surface.

Figure A.3: Ferrite orientation image of the as-welded sample from X65 pipe
with 8.6 mm wall-thickness. The image is continuous from the base metal at
the top left to the middle of the weld junction at the bottom right.

149
150

(a) Near the outer surface.

Figure A.4: Continuous ferrite orientation image from the 1 mm left towards the middle of the weld junction on the sample after
post-welding heat treatment from X65 pipe with 8.6 mm wall-thickness (continued on next page).
151

(b) At mid-thickness.

Figure A.4: Continuous ferrite orientation image from the 1 mm left towards the middle of the weld junction on the sample after
post-welding heat treatment from X65 pipe with 8.6 mm wall-thickness (continued on next page).
152

(c) Near inner surface.

Figure A.4: Continuous ferrite orientation image from the 1 mm left towards the middle of the weld junction on the sample after
post-welding heat treatment from X65 pipe with 8.6 mm wall-thickness.
Bibliography

[1] K. D. Houghton. Journal of Offshore Technology, 3(2):46–48, 1995.

[2] J. Grosse-Wördemann and S. Dittrich. Circumferential welding of

high–tensile line pipe steels under severe weather conditions (trans-
Alaska pipeline and north sea projects). In A. B. Rothwell and J. M.
Gray, editors, Welding of HSLA (microalloyed) structural steels, Mate-
rials/Metalworking Technology, pages 750–771. American Society for
Metals, Ohio, 1978.

[3] J. G. Williams, C. R. Killmore, F. J. Barbaro, A. Meta, and L. Fletcher.

In Microalloying ‘95 Conference Proceedings, pages 117–138. Iron and
Steel Society, 1995.

[4] C. Yu. Tube International, pages 153–155, March 1996.

[5] E. Treiss. 3R International, 20(11):627–630, 1981.

[6] R. R. Preston. A review of high strength, low alloy steel metallurgy

in Europe. In A. B. Rothwell and J. M. Gray, editors, Welding of
HSLA(microalloyed) structural steels, Materials/Metalworking Tech-
nology, pages 24–25. American Society for Metals, Ohio, 1978.

[7] E. Runte. The Brown Boveri Review, 55(3):113–118, 1968.

[8] W. H. Kearns, editor. High Frequency Welding, volume 3. American

Welding Society, Florida, USA, 7th. edition, 1980.

[9] S. Ramo, J. R. Whinnery, and V. Duzer. Fields and Waves in Com-

munication Electronics. John Wiley & Sons, 1984.

[10] L. F. Warren. Tube International, pages 184–186, May/June 2001.

[11] P. F. Scott and W. Smith. Tube International, pages 147–152, March

1996.

153
[12] Ö. E. Güngör, P. Yan, P. Thibaux, M. Liebeherr, H. K. D. H.
Bhadeshia, and D. Quidort. In Proceedings of the 8th international
pipeline conference, number IPC2010-31372, Calgary, Alberta, Canada,
27th Sep.–1st Oct. 2010.

[13] E. Yokoyama, A. Ejima, S. Watanabe, Y. Yoshimoto, and Y. Hirano.

Effects of welding conditions and Mn/Si ratio on the penetrator de-
fect occurrence in ERW high manganese line pipe. Technical report,
Kawasaki Steel Corporation, September 1979.

[14] J. H. Choi, Y. S. Chang, C. M. Kim, J. S. Oh, and Y. S. Kim. Welding

Journal, 83(1):27s–31s, 2004.

[15] M. D. Chapetti, J. L. Otegui, and J. Motylickl. Int. Journal of Fatigue,

24(1):21–28, 2002.

[16] J. F. Kiefner. In API’s 51st Annual Pipeline Conference & Cybernet-

ics Symp., pages 1–13, New Orleans, LA, 2000. American Petroleum
Institute.

[17] G. R. Yoder, T. W. Crooker, and L. A. Cooly. Journal of Eng. Mat.

and Tech., 101(1):86–90, 1979.

[18] H. Suzuki and A. J. McEvily. Metall. Trans. A, 10A(4):475–481, 1979.

[19] C. M. Kim, J. K. Kim, and C. S. Kim. Key Engineering Materials,

297–300:3–9, 2005.

[20] M. Beller and E. Holden. Pipes & Pipeline International, 46(6):26–34,

2001.

[21] N. Pradhan, N. Banerjee, B. B. Reddy, S. K. Sahay, D. S. Basu,

P. K. Bhor, S. Das, and S. Bhattyacharya. Scandinavian J. of Metall.,
34(4):232–240, 2005.

[22] I. V. Zakharova, E. A. Chichkarev, V. G. Vasiliev, A. I. Trotsan, A. Ya.

Dejneka, and O. S. Kiryukhin. The Paton Welding Journal, pages 14–
17, August 2001.

[23] R. W. K. Honeycombe and H. K. D. H. Bhadeshia. Steel. Edward

Arnold, London, 2nd edition, 1995.

154
[24] K. Easterling. Introduction to the Physical Metallurgy of Welding.
Butterworth-Heinemann, 2nd edition, 1992.

[25] Guide to the Welding and Weldability of C-Mn Steels and C-Mn Mi-
croalloyed Steels. Publication Document IIS/IIW-382-71. International
Institute of Welding, 1971.

[26] J. G. Williams, C. R. Killmore, F. J. Barbaro, J. Piper, and Fletcher.

Mater. Forum, 20:13–28, 1996.

[27] M. C. Zhao, K. Yang, and Y. Shan. Materials Science and Engineering

A, 335(1–2):14–20, 2002.

[28] A. Bakkaloǧlu. Materials Letters, 56(3):263–272, 2002.

[29] J. Q. Wang, A. Atrens, D. R. Cousens, and N. Kinaev. J. Mater. Sci.,

34(8):1721–1728, 1999.

[30] M. Militzer, R. Pandi, and E.B. Hawbolt. Metall. Trans. A,

27A(6):1547–1556, 1996.

[31] R. G. Kamat, E. B. Hawbolt, L. C. Brown, and J. K. Brimacombe.

Metall. Trans. A, 23A(9):2469–2480, 1992.

[32] R. A. Vandermeer. Acta Metall. Mater., 38(12):2461–2470, 1990.

[33] J. R. Bradley and H. I. Aaronson. Metall. Trans. A, 12A(10):1729–

1741, 1981.

[34] S. P. Gupta. Steel Res., 64(12):623–629, 1993.

[35] M. Avrami. J. Chem. Phys., 7(12):1103–1112, 1939.

[36] E. B. Hawbolt, B. Chau, and J. K. Brimacombe. Metall. Trans. A,

14A(9):1803–1815, 1983.

[37] P. C. Campbell, E. B. Hawbolt, and J. K. Brimacombe. Metall. Trans.

A, 22A(11):2779–2790, 1991.

[38] I. Tamura. Trans. ISIJ, 27(10):763–779, 1987.

[39] M. Enomoto and H. I. Aaronson. Metall. Trans. A, 17A(8):1381–1384,

1986.

155
[40] M. Enomoto and H. I. Aaronson. Metall. Trans. A, 17A(8):1385–1397,
1986.

[41] M. Enomoto, W. F. Lange III, and H. I. Aaronson. Metall. Trans. A,

17A(8):1399–1407, 1986.

[42] W. F. Lange III, M. Enomoto, and H. I. Aaronson. Metall. Trans. A,

19A(3):427–440, 1988.

[43] T. Gladman. The physical metallurgy of micralloyed steels. Maney,

London, 2002.

[44] J. Bauer, P. Flüss, E. Amoris, and V. Schwinn. Ironmaking and Steel-

making, 32(4):325–330, 2005.

[45] J. D. Verhoeven. J. Mater. Eng. Performance, 9(3):286–296, 2000.

[46] R. Grossterlinden, R. Kawalla, U. Lotter, and H. Pircher. Steel Res.,

63(8):331–336, 1992.

[47] E. R. Johnson and W. J. Buechling. Trans. ASM, pages 249–256,

March 1934.

[48] A. S. Bor. ISIJ Int., 31(12):1445–1446, 1991.

[49] W. C. Lyons and G. J. Plisga. Standard handbook of petroleum and

natural gas engineering. Gulf Professional Publishing, 2nd edition,
2005.

[50] Carbon and low-alloy steels. In ASM Handbook, volume 1, Proper-

ties and Selection: Irons, Steels and High Performance Alloys. ASM
International, 2002.

[51] G. Domizzi, G. Anteri, and J. Ovejero-Garcı́a. Corrosion Sci.,

43(2):325–339, 2001.

[52] Y. Nakaj, H. Kurahashi, T. Emi, and O. Haida. Development of steels

resistant to hydrogen induced cracking in wet hydrogen sulfide envi-
ronment. Technical report, Kawasaki Steel Corporation, September
1980.

[53] S. Kyriakides and E. Corona. Mechanics of offshore pipelines, volume

1. Buckling and collapse. Elsevier, Oxford, UK, 1st edition, 2007.

156
[54] SAE J310: high strength low alloy steel. In SAE Handbook, volume 1,
Materials. Society of Automotive Engineers, 1989.

[55] Fracture toughness and fracture mechanics. In ASM Handbook, volume

8: mechanical testing and evaluation. ASM International, 2002.

[56] A. H. Cottrell. Trans. TMS-AIME, pages 192–203, April 1958.

[57] N. J. Petch. Fracture. Wiley, New York, 1959.

[58] N. S. Stoloff. Fracture, volume VI: fracture of metals, chapter 1: effects

of alloying on fracture Characteristics. Academic Press, New York,
1969.

[59] D. Lonsdale and P. E. J. Flewitt. Metall. Trans. A, 9A(11):1619–1623,

1978.

[60] P. Brozzo, G. Buzzichelli, A. Mascanzoni, and M. Mirabile. Metal

Science, pages 123–129, April 1977.

[61] A. F. Gourgues, H. M. Flower, and T. C. Lindley. Mater. Sci. Technol.,

16(1):26–40, 2000.

[62] J. R. Yang, C. Y. Huang, C. F. Huang, and J. H. Aoh. J. Mater. Sci.

Lett., 12(16):1290–1293, 1993.

[63] E. Bouyne, H. M. Flower, T. C. Lindley, and A. Pineau. Scripta Mater.,

39(3):295–300, 1998.

[64] F. B. Pickering. In Proc. Symp. Transformation and hardenability in

steels, pages 109–129, Ann Harnor, MI, 1967. Climax Molybdenum Co.
and the University of Michigan.

[65] M. C. Kim, Y. J. Oh, and J. H. Young. Scripta Mater., 43(3):205–211,

2000.

[66] D. Bhattacharjee, J. F. Knott, and C .L. Davis. Metall. Mater. Trans.

A, 35A(1):121–130, 2004.

[67] D. E. McRobie and J. F. Knott. Mater. Sci. Technol., 1(5):357–365,

1985.

[68] D. Frano̧ois and A. Pineau. From Charpy to present impact testing.

Elsevier, 2002.

157
[69] F. B. Pickering. In R. W. Cahn, P. Haasen, and E. J. Kramer, ed-
itors, Materials Science and Technology, volume 7: constitution and
properties of steels. VCH, 1992.

[70] A. A. Griffith. Philos. Trans. R. Soc. A, pages 163–198, 1921.

[71] J. H. Tweed and J. F. Knott. Metal Science, 17(2):45–54, 1983.

[72] Ø. Grong, A. O. Kluken, and B. Bjørnbakk. Joining and Materials,

1:164–169, 1988.

[73] R. Kiessling. Non-metallic inclusions in steel, Part V. The Institute

of Metals, London, 1989.

[74] V. M. Radhakrishnan. Welding technology and design. New Age Inter-

national, 2007.

[75] Ø. Grong and A. O. Kluken. Key Engineering Materials, 69–70:21–46,

1992.

[76] C. L. Davis and J. E. King. Mater. Sci. Technol., 9(1):8–15, 1993.

[77] F. Matsuda, Y. Fukada, H. Okada, C. Shiga, K. Ikeuchi, Y. Horii,

T. Shiwaku, and S. Suzuki. Welding in the World, 37(3):134–154,
1996.

[78] K. J. Irvine and F. B. Pickering. J. Iron Steel Inst., pages 518–531,

June 1963.

[79] T. Gladman, B. Holmes, and I. D. McIvor. In Effect of Second-Phase

Particles on the Mechanical Properties of Steel, pages 68–78. The Iron
and Steel Institute, 1971.

[80] B. Mintz, W. B. Morrison, and A. Jones. Metals Technology, pages

252–260, July 1979.

[81] G. E. Dieter. Mechanical Metallurgy. McGraw–Hill, UK, 1988.

[82] B. L. Bramfitt and A. R. Marder. Processing and Properties of Low

Carbon Steel. AIME, New York, 1973.

[83] G. F. Melloy and J. D. Dennison. In The Microstructure and Design

of Alloys, pages 60–64, Cambridge, England, 1973. The Institute of
Metals and The Iron and Steel Institute.

158
 [84] R. L. Plaut. PhD thesis, Sheffield University, UK, 1974.

[85] W. B. Morrison. In Controlled Processing of HSLA Steels. Sheffiled:

British Steel Corperation, 1976.

[86] W. R. Tyson, R. A. Ayres, and D. F. Stein. Acta Metallurgica,

21(5):621–627, 1973.

[87] V. M. Goritskii and D. P. Khromov. Problemy Prochnosti, pages 81–82,

June 1984.

[88] G. J. Baczynski, J. J. Jonas, and L. E. Collins. Metall. Mater. Trans.

A, 30A(12):3045–3054, 1999.

[89] J. W. Christian. Metall. Trans. A, 14A(7):1237–1256, 1983.

[90] J. Ju, J. Lee, and J. Jang. Materials Letters, 61(29):5178–5180, 2007.

[91] H. Yu. Journal of University of Science and Technology Beijing, Min-

eral, Metallurgy, Material, 15(6):683–687, 2008.

[92] R. Rittmann and K. Freier. In Niobium Science and Technology: Proc.

International Symp. Niobium 2001, Orlando, Florida, USA.

[93] D. A. Hughes and N. Hansen. Acta Metallurgica, 45(9):3871–3886,

1997.

[94] S. Mironov, Y. S. Sato, and H. Kokawa. Acta Materialia, 56(11):2602–

2614, 2008.

[95] S. Li, I. J. Beyerlein, and M. A. M. Bourke. Materials Science and

Engineering A, 394(1–2):66–77, 2005.

[96] S. Kundu. PhD thesis, University of Cambridge, 2007.

[97] S. T. Kimmins and D. J. Gooch. Metal Science, 17(11):519–532, 1983.

[98] S. Matsuda and Y. Okamura. Trans. Iron Steel Inst. Jpn., 14(1):444–
449, 1974.

[99] S. Matsuda and Y. Okamura. Trans. Iron Steel Inst. Jpn., 14(1):363–
368, 1974.

[100] R. Homma. Trans. Iron Steel Inst. Jpn., 14(1):434–443, 1974.

159
[101] J. Hilliard. Metal Progress, pages 99–100, May 1964.

[102] P. J. Goodhew, J. Humphreys, and R. Beanland. Electron microscopy

and analysis. Taylor & Francis, 3rd. edition, 2001.

[103] Oxford Instruments. Manual of HKL Technology Channel 5, 2007.

[104] J. R. Michael. In N. Yao and Z. L. Wang, editors, Handbook of

microscopy for nanotechnology, chapter 13: characterization of nano-
crystalline materials using electron backscatter diffraction in the scan-
ning electron microscope. Kluwer Academic Publishers, New York,
2005.

[105] S. X. Ren, E. A. Kenik, K. B. Alexander, and A. Goyal. Microsc.

Microanal., 4(1):15–22, 1998.

[106] S. Kikuchi. Proc. Imp. Acad., 4(6):271–278, June 1928.

[107] P. V. C. Hough. Method and means for recognizing complex patterns.

US patent 3,069,654, 1962.

[108] D. J. Dingley. In Electron backscatter diffraction in materials science,

chapter 1: the development of automated diffraction in scanning and
transmission electron microscopy. Kluwer Academic Publishers, New
York, 2000.

[109] G. S. Rohrer. Structure and bonding in crystalline materials. Cam-

bridge University Press, 2001.

[110] B. L. Adams, S. I. Wright, and K. Kunze. Metall. Trans. A,

24A(4):819–831, 1993.

[111] J. R. Michael. In Electron backscatter diffraction in materials science,

chapter 7: phase identification using electron backscatter diffraction in
the scanning electron microscope. Kluwer Academic Publishers, New
York, 2000.

[112] Y. Li, X. Han, L. Ji, Y. Feng, and C. Huo. In Proceedings of 7th

international pipeline conference, Calgary, Alberta, Canada, 2008.

[113] V. Venegas, F. Caleyo, J. L. González, T. Baudin, J. M. Hallen, and

R. Penelle. Scripta Mater., 52(2):147–152, 2005.

160
[114] V. Venegas, F. Caleyo, J. M. Hallen, T. Baudin, and R. Penelle. Metall.
Mater. Trans. A, 38A(5):1022–1031, 2007.

[115] L. S. Birks. In P. J. Elving and I. M. Kolthoff, editors, Electron probe

microanalysis, volume 17 of Chemical analysis. Wiley-Interscience,
1963.

[116] G. R. Speich, L. J. Cuddy, C. R. Gordon, and A. J. DeArdo. In A. R.

Marder and J. I. Goldstein, editors, Phase Transformations in Ferrous
Alloys, pages 341–389, Warrendale, Pennsylvania, USA, 1984. TMS–
AIME.

[117] R. L. Bodnar and S. S. Hansen. Metall. Mater. Trans. A, 25A(4):665–

675, 1994.

[118] S. Jones and H. K. D. H. Bhadeshia. Metall. Mater. Trans. A,

28A(10):2005–2103, 1997.

[119] S. Jones and H. K. D. H. Bhadeshia. Acta Materialia, 45(7):2911–2920,

1997.

[120] H. Kristoffersen and P. Vomacka. Materials and Design, 22(8):637–644,

2001.

[121] K. Fujita, A. Yoshida, and K. Nakase. Bull. JSME, 22(169):994–1000,

1979.

[122] A. Lambert-Perlade, A. F. Gourgues, and A. Pineau. Acta Materialia,

52(8):2337–2348, 2004.

[123] Y. M. Kim, S. Y. Shu, H. Lee, B. Hwang, S. Lee, and N. J. Kim. Metall.

Mater. Trans. A, 38A(8):1731–1742, 2007.

[124] S. Y. Shin, S. Y. Han, B. Hwang, C. G. Lee, and S. Lee. Materials

Science and Engineering A, 517(1–2):212–218, 2009.

[125] Y. Li and T. N. Baker. Mater. Sci. Technol., 26(9):1029–1040, 2010.

[126] J. H. Chen, Y. Kituta, T. Araki, M. Yoneda, and Y. Matsuda. Acta

Metallurgica, 32(10):1779–1788, 1984.

[127] C. Wang, X. Wu, J. Liu, and N. Xu. Materials Science and Engineering
A, 438–440:267–271, 2006.

161
[128] S. Lee, B. C. Kim, and D. Kwon. Metall. Trans. A, 23A(10):2803–2816,
1992.

[129] National Physical Laboratory. NPL MTDATA software, 2006. Ted-

dington, UK.

[130] G. Krauss and S. W. Thompson. ISIJ Int., 35(8):937–945, 1995.

[131] Y. E. Smith, A. P. Coldren, and R. L. Cryderman. In Proc. Symp.

Toward improved ductility and toughness, pages 119–142, Ann Harnor,
MI, 1971. Climax Molybdenum Co.

[132] S. W. Thompson, D. J. Colvin, and G. Krauss. Metall. Trans. A,

21A(6):1493–1507, 1990.

[133] Atlas for bainitic microstructures, volume 1. ISIJ, Tokyo, 1992.

[134] L. J. Habraken and M. Economopoulos. In Proc. Symp. Transforma-

tion and hardenability in steels, pages 69–107, Ann Harnor, MI, 1967.
Climax Molybdenum Co. and the University of Michigan.

[135] D. J. C. MacKay. In H. Cerjak, editor, Mathematical Modelling of Weld

Phenomena, volume 3, pages 359–389. Institute of Materials, Minerals
and Mining, 1997.

[136] R. C. Eberhart and R. W. Dobbins. Neural Network PC Tools A prac-

tical guide. Academic Press, California, 1990.

162